<%BANNER%>

Characterization of the University of Florida air-water shear-layer facility

University of Florida Institutional Repository

PAGE 1

i CHARACTERIZATION OF THE UNIVERSITY OF FLORIDA AIR-WATER SHEARLAYER FACILITY By SRIHARI RASMI A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2002

PAGE 2

ii ACKNOWLEDGMENTS I would like to thank my advisor, Dr. Louis Cattafesta, for his constant encouragement and guidance. I also thank Dr. Mark Sheplak for his support and tutelage efforts. Mr. Kurt Banazynski of ELD Inc. was instrumental in the manufacturing process of the University of Florida air-water shear layer facility. I also thank him for his promptness in attending to matters related to the facility. I greatly acknowledge the presence of Chris Bahr who helped me out in setting up and carrying out various experiments. I also thank Ryan Holman for kindly letting me borrow his codes to work upon. Financial support for this project was provided by the Office of Naval Research under Grant number N0014-00-1-0105. ii

PAGE 3

iii TABLE OF CONTENTS Page ACKNOWLEDGMENTS...................................................................................................ii LIST OF TABLES..............................................................................................................v LIST OF FIGURES............................................................................................................vi ABSTRACT....................................................................................................................... ix CHAPTERS 1 INTRODUCTION..........................................................................................................1 Supercavitation................................................................................................................ 3 Motivation for Facility....................................................................................................6 2 DESIGN AND CONSTRUCTION..............................................................................11 Six-inch Flow Visualization Water Tunnel...................................................................11 Flow Sections............................................................................................................12 Test Section...............................................................................................................13 Pump……..................................................................................................................15 Variable Speed Drive Assembly...............................................................................15 Piping and Supporting Framework...........................................................................15 Six-inch High-Speed, Open-Circuit Wind Tunnel........................................................16 Flow Ducts and Flow Straighteners..........................................................................17 Test Section...............................................................................................................17 Primary Diffuser and Liquid-Gas Separator.............................................................19 Fan Assembly............................................................................................................20 Motor and Motor Controller......................................................................................20 Supporting Frame......................................................................................................20 3 CHARACTERIZATION OF FACILITY.....................................................................22 Velocity Profiles of the Air Tunnel...............................................................................22 Experimental Set-up..................................................................................................22 Calibration of Air Tunnel Velocity and Air Pressure Transducer.............................24 Velocity Profiles........................................................................................................25 Velocity Profiles of the Water Tunnel..........................................................................27 Experimental Set-up..................................................................................................27 Calibration of Water Tunnel Velocity and Water Pressure Transducer....................28 Velocity Profiles........................................................................................................29 iii

PAGE 4

iv Vibration Measurements of the Test Section................................................................31 Experimental Set-up..................................................................................................31 Theoretical Model.....................................................................................................33 Results and Conclusions............................................................................................36 4 INTERFACE EDUCTION USING LIGHT SHEET...................................................44 Experimental Set-up......................................................................................................44 Procedure...................................................................................................................45 Interface Detection....................................................................................................47 Outlier Rejection.......................................................................................................49 Transformation and Calibration................................................................................51 Interface Slope Calculation.......................................................................................54 CONCLUSIONS AND FUTURE WORK.......................................................................56 APPENDIX TUNNEL OPERATION PROCEDURE..........................................................................57 REFERENCES..................................................................................................................60 BIOGRAPHICAL SKETCH............................................................................................62 iv

PAGE 5

v LIST OF TABLES Table Page 3-1 Air tunnel velocity characteristics.............................................................................27 3-2 Water tunnel velocity characteristics.........................................................................31 3-3 Test matrix for vibration measurement......................................................................33 4-1 Critical parameters of the laser and camera...............................................................45 4-2 Test matrix for interface eduction..............................................................................46 4-3 The maximum error between the data and polynomial fit.........................................54 v

PAGE 6

vi LIST OF FIGURES Figure Page 1-1 Tip vortex cavitation in a marine propeller..................................................................2 1-2 Schematic of a supercavitating vehicle........................................................................3 1-3 Schematic of interfacial conditions in the air-water shear layer facility.....................6 1-4 Schematic of interfacial conditions of a supercavitating vehicle.................................7 1-5 Air-water test facility for sensor characterization........................................................8 1-6 Velocity profiles of a supercavitating vehicle (left) and the air water shear layer facility (right)..............................................................................................................9 2-1 Inlet plenum and pump..............................................................................................13 2-2 Return plenum and test section..................................................................................14 2-3 Test section of University of Florida shear layer facility..........................................14 2-4 First part blower housing, wide-angle diffuser, flow straighteners and contraction section........................................................................................................................ 16 2-5 Primary diffuser and liquid gas separator..................................................................19 2-6 Rubber-in-shear mounts to minimize vibrations........................................................21 3-1 Experimental set-up used to measure velocity profile in air tunnel...........................23 3-2 Schematic of the pitot-static probe mounting in the test section...............................23 3-3 Motor frequency and velocity of air tunnel plotted as a function of DAQ input voltage.......................................................................................................................2 5 3-4 Block diagram for calibration and velocity profile measurement for the air tunnel..25 vi

PAGE 7

vii 3-5 Velocity profiles (with error bars) of the air tunnel for velocities increasing from 6 m/s to 66 m/s in ten steps (corresponding to motor input voltage increasing from 1 V to 10 V in steps 1 V)..................................................................................................26 3-6 Motor frequency of water tunnel plotted as a function of DAQ input voltage..........29 3-7 Block diagram for calibration and velocity measurement for the water tunnel.........29 3-8 Velocity profiles (with error bars) of water tunnel for velocity increasing from 0.15 m/s to 0.86 m/s in seven steps (corresponding to motor input voltage increasing from 1 V to 8 V in steps of 1 V)........................................................................................30 3-9 Accelerometers mounted at possible vibration sources, i.e. at the (a) blower and the (b) pump respectively................................................................................................32 3-10 Accelerometer mounted approximately at the midpoint of the test section to detect vibration levels. Flow direction is along Y..............................................................32 3-11 Power spectra of three components for all three accelerometers.............................34 3-12 Two-input single output model for vibration analysis.............................................34 3-13 Top two plots show the ordinary coherent output spectrum due to the inputs. The bottom two plots compare the noise spectrum with the output spectrum, the multiple coherent output spectrum and the noise floor...........................................................38 3-14 The topmost plot shows 2 12 the coherence between the inputs. The next two plots are 2 1 y and 2 2 y the ordinary coherence functions between each input and the output. The bottommost plot is 2 : y x the multiple coherence function..................................39 3-15 Comparison of output spectrum with BK microphone measurements. In the top plot, the microphone is placed very close to the blower, with the pump off. In the bottom plot, the microphone is placed very close to the pump, with the blower off.41 3-16 Comparison of the output spectrum with microphone measurements for the case when the fan is off and the water tunnel speed is 0.5 m/s.........................................42 4-1 Experimental set-up with Nd:YAG LASER and PIVCAM......................................45 4-2 Sample original image (left) and corresponding output from edge detection algorithm (right). The flow direction is from left to right as indicated....................................47 4-3 A sample edge image and the corresponding interface image as a result of pruning.48 vii

PAGE 8

viii 4-4 Average interface and standard deviation of air-water interface. Number of statistics: 200............................................................................................................................ .49 4-5 Superposition of sequence of edge images before outlier rejection and a column (at 188th X pixel) of sum of sequence of interface images with an outlier point..........49 4-6 Sum of sequence of edge images after outlier rejection and a column (at 188th X pixel) of sum of sequence of interface images without any outlier points................50 4-7 Comparison of interface images without and with outlier rejection. The average images are shown before outlier rejection (left) and after outlier rejection (right)...51 4-8 Comparison of standard deviation values before and after outlier rejection.............51 4-9 Projection transformation and reverse projection transformation of the calibration image.........................................................................................................................5 2 4-10 The average interface image and the corresponding reverse transformation to world coordinates................................................................................................................53 4-11 Interface image in world coordinates and corresponding polynomial fit................54 4-12 Polynomial fit, of order 5, of average interface image (left) and corresponding slope of interface (right).....................................................................................................55 viii

PAGE 9

ix Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science CHARACTERIZATION OF THE UNIVERSITY OF FLORIDA AIR-WATER SHEAR LAYER FACILITY By Srihari Rasmi December 2002 Chairman: Dr. Louis N. Cattafesta Major Department: Mechanical and Aerospace Engineering Technological advancements in many fields have contributed to envisioning supercavitating vehicles that can travel at far greater speeds than current underwater vehicles. Under supercavitating conditions, the vehicle is enveloped in a single gaseous bubble also known as a supercavity. Realizing such a vehicle admits many possibilities for research in related fields. One of the chief issues is monitoring the cavity thickness to control the vehicle. For this purpose, new algorithms and sensors have been developed apart from performing comprehensive numerical simulations, etc. For testing the sensors and algorithms, a novel facility has been designed and constructed. The facility is an air-water shear-layer system that consists of an air and a water test section with a free surface. The integral test section is 48 in. long by 6 in. wide by 12 in. high. The air stream speed can reach a maximum of approximately 75 m/s, corresponding to the approximate speed of a supercavitating vehicle. The water section is designed with a maximum speed of approximately 0.8 m/s, which is high enough to ix

PAGE 10

x produce natural cavitation bubbles. The interfacial instabilities arising at the free surface allow for sensor characterization and algorithm testing in an environment that simulates supercavitation while maintaining low cost. In this thesis, some of the more important features such as the velocity profiles of air and water streams, vibration characteristics and interface conditions are studied. Velocity profile measurements of both air and water tunnels are carried out using pitotstatic probes. The velocity profiles of the air tunnel are approximately uniform throughout the test section outside of the boundary layer. For the water tunnel, large error bars lead to inconclusive results regarding the mean flow quality in the test section. Vibration measurements are made using accelerometers. This is to deduce the influence of vibration from fan and pump (driving the air and water tunnels) on the flow quality needs as a part of future work. An interface eduction method is developed to extract the interface from series of flow visualization images for a particular interface condition. Statistical quantities related to interface shape are obtained and discussed. x

PAGE 11

1 CHAPTER 1 INTRODUCTION Realizing high speeds in underwater vehicles has been a challenge mainly because of propulsion limitations and skin friction drag. The problem of skin friction drag is compounded by the fact that it increases with the square of the velocity of the vehicle. However, the last half-century has seen significant advancement in technological concepts that makes achievement of greater speeds a definite possibility for underwater vehicles [Tulin 1961]. Such a possibility has been envisioned as a result of a phenomenon known as supercavitation. Under supercavitating conditions, an underwater vehicle is entirely encapsulated in a large gaseous bubble or cavity. Realizing this technology requires a thorough understanding of the physics of the problem. In addition, control and maneuverability of the vehicle pose new challenges because of the different dynamics involved. In this situation, the measurement of the cavity shape with appropriate sensors is imperative. The characterization of these sensors demands an innovative test bed that is cost effective and, at the same time, able to simulate the supercavitation environment to a fair degree. This thesis aims at summarizing the development, design and characterization of such a test bed for sensor testing specific to supercavitating vehicles. The need for a unique test facility is built up from the definition of cavitation as given below. Cavitation is defined as the gas-liquid region formed due to pressure reduction as a result of the dynamic action of the fluid on the boundaries of a liquid system [Stutz and 1

PAGE 12

2 Reboud 1997]. This pressure reduction can be roughly quantified in the following way. Bernoulli's equation along a streamline with negligible height change is 21 vconstant 2 p (1.1) where p is the pressure, is the density and v is the speed of the liquid. When v increases sufficiently so that p drops below the vapor pressure v p of the liquid, the formation of bubbles, also known as cavities that contain the vapor of the liquid occurs. This is cavitation. Based on this, the “cavitation number” is defined as 22 vv p p (1.2) The cavity can be filled with the vapor from the ambient liquid, as in the case of the water vapor bubbles at the tip of a propeller blade in tip-vortex cavitation shown in Figure 1-1 [Billet 2000]. However, a cavity can also be filled with any other gas. For example, the formation of cavities of air in the rear of any body during water entry is very common. Albeit the latter case does not adhere to definition of cavitation, the physics is similar. Hence, such cavities are referred to as ventilated cavities [Tulin 1961]. Figure 1-1 Tip vortex cavitation in a marine propeller (from Billet 2000).

PAGE 13

3 Cavitation is usually very localized and for the most part is known to have detrimental effects. For example, in bubble cavitation of marine propellers, the rapid formation and collapse of bubbles can be very harmful to the blade of the propeller [Stutz and Reboud 1997]. As a result the blade suffers from fatigue and loss of material. Hence a great incentive lies in predicting cavitation and studying the effect of cavitation with respect to marine propellers. In general, cavitation occurs in various environments like low head pumps, heart valves, venturi meters etc. To model these, analytical methods that employ potential theory have been used with a degree of success [Kunz et al. 2000]. However, cavitation often includes mass transfer, unsteadiness and viscous effects [Stutz and Reboud 1997; Kunz et al. 2000]. Supercavitation Supercavitation is a specific cavitating condition wherein a vehicle moving in a liquid is completely enveloped by a gaseous cavity. In a supercavitating condition, only a single gaseous bubble exists that is usually much larger than the body itself and stays attached to the body at all times [Tulin 1961]. A schematic of a supercavitating vehicle is shown in Figure 1-2 [Kam Ng 1999]. Figure 1-2 Schematic of a supercavitating vehicle [Kam Ng 1999]. The incentive to employ supercavitation in underwater vehicles is that the skin friction drag is reduced to a great extent because the viscosity and density of the gas is Supercavity Cavitato r

PAGE 14

4 considerably less than that of the liquid [Werner 1998]. This leads to greater thrust and thus greater speeds. High-speed supercavitating vehicles can travel at hundreds of knots [Tulin 1961; Kulkarni and Pratap 1999]. Supercavitation can take two forms. First, natural cavitation occurs when the vehicle is traveling fast enough to vaporize the ambient liquid and form a single bubble cavity. Second, artificial cavitation or ventilation occurs when a ventilator is present at the nose of the body to inject gases and form a cavity (Figure 1-2). In the second method, which is more common, the exhaust gases are fed through the ventilator. Irrespective of natural or artificial inception of the cavity, the term supercavitation is applied for both cases [Tulin 1961]. The optimal design of a supercavitating vehicle results in significant drag reduction, since only small regions at the nose and afterbody control surfaces contact the water. Design considerations in such vehicles mainly include the shape of the cavitator, vehicle size and shape, shape of fins, and dynamics of the vehicle. The cavitator dictates the shape and size of the cavity. Cavitator shapes in the form of disks, cones, etc. have been experimented with [Stinebring et al. 2000, 2001]. The small size of the cavitator with respect to the body ensures minimal pressure drag for the speed range of interest. With the size of the cavity fixed by the shape and orientation of the cavitator and the given speed range, the size of the vehicle itself is restricted as a result of this. The vehicle houses the propulsion system and control and guidance system among other parts. The fins are the primary control surfaces that mostly cut well into the cavity wall (Figure 1-2). The dynamics of the problem are increasingly complicated when the vehicle turns

PAGE 15

5 wherein tail slap at the cavity wall is a definite possibility [Kulkarni and Pratap 1999]. Vehicle rotation has also been predicted [Kulkarni and Pratap 1999]. The aforementioned design considerations and conditions have led to piecewise research in specific areas like cavitator designs [Stinebring et al. 2001], numerical prediction [Kunz et al. 2000], cavity thickness measurement using novel methods [Li et al. 2002a, 2002b, 2002c] etc. However, supercavitation itself has been studied for a long time. The use of supercavitation in high-speed marine propellers, hydrofoil boats, and low-head pumps/turbines sparked a renewed interest in the subject [Tulin 1961]. Added to this, supercavitation with respect to bodies of revolution have been under scrutiny in recent times especially with regard to underwater artillery [Tulin 1961]. As a result, associated topics such as water entry, dynamics, and control have been the subjects of theoretical and experimental studies [Tulin 1961]. While supercavitation applied to underwater vehicles is of a great tactical merit, one of the principal tradeoffs is the stability and maneuverability of the vehicle. While the stability is dictated by the design of the vehicle and also by the dynamics of the cavity, the maneuverability depends on the prediction and control of the cavity shape and vehicle dynamics in the presence of fins. For the prediction and control of the cavity, there is a need to (a) measure the local cavity thickness, (b) understand the interfacial instabilities at the cavity wall and (c) employ a suitable control algorithm. The measurement of cavity thickness has necessitated the need for innovative technology. Related research areas including algorithm development, sensor manufacturing and experimentation are in progress [Chandrasekharan et al. 2001a, 2001b, 2002; Li et al. 2002a, 2002b, 2002c].

PAGE 16

6 Motivation for Facility For the new algorithms and sensors that are currently being developed, a cost efficient testbed that mimics supercavitating conditions is required for experimentation. This provides the primary motivation for the University of Florida air-water shear-layer facility. This is because setting up a conventional supercavitating environment in lab conditions is a challenging and costly problem. Such a set-up would require a test section in which water flows at high speeds, even to maintain a supercavity under ventilating conditions. To circumvent the complexity and cost factor involved in such a project, an affordable solution is sought in the form of an air-water shear layer interface (see Figure 1-3). A supercavitating environment is mimicked by keeping the air speed at approximately 150 knots and maintaining the speed of water to be high enough to produce cavitation bubbles (approximately 1 m/s) [Sheplak et al. 1999]. This system is certainly a low cost option as conventional low speed air tunnels and water tunnels can be manufactured at a much lower cost than a custom water tunnel that has to propel water at high speeds. Figure 1-3 Schematic of interfacial conditions in the air-water shear layer facility. To assess the ability of an air-water shear layer facility to simulate the actual supercavitating conditions, a comparison of the air-water interface conditions in both

PAGE 17

7 cases is now addressed. For a supercavitating vehicle, a schematic of the cavity interface is shown in Figure 1-4. Figure 1-4 Schematic of interfacial conditions of a supercavitating vehicle. The various stages of the interface along the length of the underwater vehicle are illustrated. Interfacial instabilities that are initiated at the nose of the vehicle propagate towards the aft of the vehicle, growing in amplitude. Hence, while the cavity near the nose of the vehicle is minimally unsteady, the closing region of the cavity suffers considerable interfacial unsteadiness. At the nose, the cavity surface is more or less glasslike. Typically very few gas bubbles exist in this region. The next stage shown indicates the amplification of instabilities initiated at the nose of the vehicle. Near the aft of the vehicle, the air-water interface may break down, leading to very high drag forces. This region may contain high concentrations of water vapor. Also, there is significant mixing of the two fluids. For comparison, a schematic of the air-water interface of the air-water shear layer facility is shown in Figure 1-3. A splitter plate separates the air and water streams. Much like in a supercavitating vehicle, the instabilities begin at the splitter plate and propagate along the flow direction. The region near the splitter plate

PAGE 18

8 suffers very small disturbances whereas towards far downstream of the test section, amplification results in significant disturbances. The shear-layer facility is established as a competent substitute to actual supercavitating conditions. To test distance estimation algorithm and sensors, a sensor package housed in a flat plate is inserted in the air stream of the air-water shear layer facility as shown in Figure 1-5. A flat plate that houses the sensor package with an embedded sensor package is inserted into the high-speed air stream. Case studies, by varying the position x the orientation and height h of the sensor above the interface, are possible for various conditions of the air/water interface. Figure 1-5 Air-water test facility for sensor characterization. The main drawback of the air-water shear-layer facility is that it does not exactly match the operating conditions of a high-speed supercavitating vehicle. The difference arises from the non-moving flat plate, which leads to different velocity profiles between the two cases. This is shown in Figure 1-6. With the supercavitating vehicle traveling at high speed, the gas adjacent to it is at the same speed in accordance with the no-slip

PAGE 19

9 boundary condition. Hence, the velocity monotonically asymptotes to zero while maintaining continuity at the interface. Whereas in the air-water shear-layer facility, since the plate is stationary and the free stream speed is nonzero, a boundary layer develops on the flat plate. The state of the boundary layer that develops and its state can be controlled by proper selection of the plate length and/or a boundary-layer trip [Sheplak et al. 1999]. From the plate, the velocity increases from zero, in accordance with the no slip boundary condition, to the free stream velocity and then asymptotically drops to the water stream speed. Notwithstanding this difference, the air water shear layer facility does simulate many of the important features of the cavity interface of a supercavitating vehicle. Figure 1-6 Velocity profiles of a supercavitating vehicle (left) and the air water shear layer facility (right). The facility described in Chapter 2 is fabricated by Engineering Laboratory Design, Inc. The salient features of this facility are 1) a plexiglas test section with dimensions 6 in. wide by 12 in. high by 48 in. long (a 6 in. square section for each

PAGE 20

10 stream), 2) a maximum air speed of approximately 76 m/s, and 3) a maximum water speed at just under 1 m/s which is high enough to produce natural cavitation bubbles. The next chapter will focus on a detailed design of the air water shear layer facility along with description of the chief components used.

PAGE 21

11 CHAPTER 2 DESIGN AND CONSTRUCTION The University of Florida Air-Water Shear Layer Facility is a unique testbed developed for sensor characterization and algorithm testing related to supercavitating vehicles. This chapter is relies heavily on information presented in Banazynski [2000]. The facility has overall dimensions of 27 ft. long by 3.92 ft. high by 6.2 ft. wide and a net weight of 1900 lbs. The test section (Figure 2-3) measures 48 in. in length by 6 in. in width by 12 in. high. The centerline of the test section is approximately 62 in. from the ground. The facility consists of three principal parts that can be readily joined/disjoined. The first part houses the inlet, blower, contraction section, etc. The second part comprises the closed-loop water tunnel inclusive of the test section and water pump. The third part is the primary diffuser of the wind tunnel with a liquid gas separator unit. These three parts are carefully aligned and joined by screws. The shear-layer facility can be classified according to its two main components: the water tunnel and the air tunnel. The design and parts of the water tunnel are described in detail below followed by the same for the air tunnel. Six-inch Flow Visualization Water Tunnel The water tunnel is an un-pressurized system that is designed to operate in a closedcircuit configuration a vertical flow loop. It consists of the pump, flow sections, 11

PAGE 22

12 supporting framework and the test section. The test section may be operated as a conventional water tunnel with a plexiglas cover in place. This unit is 10.375 ft. long by 5.58 ft. high by 2.5 ft. wide. The test section centerline height is 4.52 ft. from the ground. The gross weight of the system, filled with water, is approximately 660 lb. The tunnel can hold approximately 48 gallons of water. The overall construction and different components used in the water tunnel are described below. Flow Sections All elements of the system in contact with water are fabricated from non-corrosive materials. The flow sections are fabricated from laminated fiberglass reinforced plastic. This material is used because it is strong and lightweight. It easily withstands both applied static and dynamic loads. Interior surfaces have a glass smooth, white, vinylester, gel-coat. Exterior surfaces are spray finished with a medium blue polyester and gel-coat enamel. To ensure water tightness, flanged joints are sealed using a high quality, polyurethane marine adhesive/sealant. All joints are secured using stainless steel fasteners. Any discontinuities in fit and alignment could lead to disturbances in flow. Hence, careful attention is paid in joining mating sections. The covers for the fiberglass ducts are fabricated from 0.5 in. thick plexiglas. The inlet plenum (Figure 2-1), also called the distributed plenum, and the return plenum are adjacent to the test section on either side. A perforated cylinder distributes flow from a centrifugal pump into the diffusing part of the inlet plenum. Perforated plates of stainless steel act as head loss baffles in the diffuser. For a settling chamber, a precision, tubular cell, plastic, honeycomb section is inserted just before the contraction.

PAGE 23

13 Adjacent to this, two stainless steel screens (60% porosity) are mounted. Both the honeycomb section and stainless steel screens act as flow straighteners. A removable cover provides access to this area. In the return plenum (Figure 2-2), twin-turning vane cascades are placed to direct the flow leaving the test section. A removable plexiglas end wall enables observation from downstream. 1.3.1 Test Section The test section sides and flanges are made of 0.5 in. thick, clear acrylic plexiglas. The test section floor is made of 0.75 in. thick, clear, acrylic plexiglas. The test section interior is 48 in. long by 6 in. high by 6 in. wide (Figure 2-3). A portion of this excess height is used as the sidewall boundary for the wind tunnel test section. Figure 2-1 Inlet plenum and pump.

PAGE 24

14 Figure 2-2 Return plenum and test section. Figure 2-3 Test section of University of Florida shear layer facility. To operate the water tunnel in a conventional water-tunnel only mode, a cover fabricated from in. thick, clear acrylic plexiglas is provided. The cover has lap flanges

PAGE 25

15 that correspond to steps machined into the permanent mid-height ends in the test section. The cover is secured to the mid-height ends with nylon screws. 1.3.2 Pump A stainless steel centrifugal pump (G&L SSH Model 6SH2F2G0, with a 5.625 in. diameter impeller) is provided to circulate the water in the tunnel at a desired speed. The pump delivers 280 GPM using 1.5 HP. The pump is directly driven by a 1.5 HP, open drip-proof, 1800 RPM, 208/230 VAC induction motor (Baldor International Model No. JMM3154T). 1.3.3 Variable Speed Drive Assembly The pump shaft RPM (frequency) is controlled by a transistor inverter type, variable frequency inverter (Toshiba Model No. VFS7S-2015UP). The inverter is arranged for 208/230 VAC, single-phase input electric service. A remote control station, located adjacent to the test section, provides a user-friendly interface to regulate the test section velocity. A NEMA rate fusible disconnect is wired inline to protect both the motor and controller. The disconnect, inverter and related components are mounted in a NEMA Type 4 enclosure. 1.3.4 Piping and Supporting Framework The majority of the piping is commercial PVC pipe and fittings. Flexible rubber couplings are used to join the piping to the flow sections to provide water tightness and to act as vibration dampers. A drain valve is provided at the lowest point in the system. The supporting framework consists primarily of structural steel tubing. The framework consists of two parts: the main frame and the pump frame. The main frame

PAGE 26

16 holds the flow section and test section. The main frame is fitted with swivel casters for transportation. The pump frame supports the pump to isolate the test section from pump vibrations. Each frame leg is fitted with an adjustable leveling pad. The pump frame is carried on an extension of the main frame. The frames are etched, prime coated and spray finished with acrylic enamel. Six-inch High-Speed, Open-Circuit Wind Tunnel The wind tunnel system is an open-circuit blower-type system. Air is sucked normal to the flow direction into the inlet by the fan as shown in Figure 2-4. The system consists of flow sections and straighteners, test section, primary diffuser and a liquid-gas separator unit. The unit is 27.2 ft. long by 6.29 ft. high by 3.45 ft. wide. The total weight of the unit is approximately 1250 lbs. Figure 2-4 First part blower housing, wide-angle diffuser, flow straighteners and contraction section. The overall construction and different components used in the wind tunnel are described below. Air intake

PAGE 27

17 1.3.5 Flow Ducts and Flow Straighteners The flow ducts are fabricated from fiber reinforced plastic molded over precision tooling. Interior surfaces have glass smooth, white, vinyl-ester, gel-coat, while exterior surfaces are spray finished with a medium blue, polyester, gel-coat enamel. The system air from the blower is discharged through a wide-angle diffuser that has perpendicular lap flanges. Then face flanges are fastened with bolts. The wide-angle diffuser section expands with a total included angle of 12.3 in the horizontal plane and 1.4 in the vertical plane. The wide-angle diffuser helps in lowering the speed before the air enters through the flow straighteners. In the settling length, a precision, hexagonal cell, aluminum honeycomb section is placed. This is followed by a mesh of high porosity (60%) stainless steel screens. The screens are mounted and tensioned using a proprietary design. Extruded aluminum frames cover the entire settling length. The settling length is followed by the contraction section that is a single-piece mold. The cross section of the contraction section is symmetric throughout, has a 9:1 ratio, and has analytically developed contours. The contraction section leads to the test section through a transition plexiglas section. Flow from the test section continues into the diffuser section that has miter flanges. The primary diffuser section expands with a total included angle of 7. High porosity perforated plates are fastened 14.81 in. from the diffuser entrance and at the diffuser exit. Any discontinuities in fit and alignment could lead to disturbances in flow. Hence, careful attention is paid in joining mating sections. 1.3.6 Test Section The test section is integral to the water tunnel test section. The water tunnel test section is 48 in. long by 6 in. high by 6 in. wide in the interior. The total test section is a

PAGE 28

18 single plexiglas mold that is 48 in. long by 12 in. high by 6 in. wide. The removable test section is joined to the plexiglas transition section and the diffuser ducts with flanges. The wind tunnel cover consists of a hinged ceiling plate that can be raised or lowered with a screw. As a result, the ceiling can be diverged from 0 to 1.8. To minimize air leak at the sides of the ceiling plate, a self-adhesive felt gasket is used as sealant. The ceiling for the duct downstream from the test section is connected to the diverging ceiling and transitions back to the 6.00 in. test section height. Along the span wise centerline of the ceiling plate, a 0.63 in. wide by 43.18 in. long slot is machined on the diverging part. The slot allows access of instrumentation into the test section. To minimize air leaks through the unused area of the slot, a high-density nylon brush with extruded aluminum holder is used. This brush is secured to the ceiling plate with screws. A series of 22 static pressure taps (with nylon barbed fittings) are located at 1.50 in. distance from the spanwise centerline towards the sidewall. The centers of the static pressure ports are placed at an interval of 2.13 in. along the flow direction. To run the test section as a conventional air wind tunnel, a removable floor, fabricated from 0.5 in. thick, clear acrylic plexiglas is provided. The floor has lap flanges that correspond to steps machined into the permanent mid-height ends in the test section. The floor is secured with screws. An additional, shorter wind tunnel test section is provided to allow the wind tunnel to be used separately from the water tunnel. The test section can be mounted directly to the contraction exit. The working section is 18 in. long by 6 in. high by 6 in. wide in the interior. Plexiglas flanges join the test section to the contraction section and the diffuser

PAGE 29

19 ducts. A 15 in. portion of the ceiling is removable and secured with quick release fasteners. 1.3.7 Primary Diffuser and Liquid-Gas Separator The primary diffuser (Figure 2-5) serves to regain static pressure. It is a single-fiber reinforced plastic mold that is joined with screws to the return plenum. The primary diffuser section expands with a total included angle of 7. Flow continues through a liquid/gas separator unit and is discharged to the atmosphere. A liquid-gas separator is necessary because operation of the facility at air speeds greater than 40 m/s results in a significant loss of water from the shear layer interface due to entrainment by the airflow. The liquid/gas separator is a commercial model Harrington Model No. HPE-30. The liquid/gas separator is installed at the farthest point downstream of the wind tunnel diffuser. The water reclaimed by the mist eliminator can be discharged via a drain or routed back to the water tunnel using a commercial pump. Figure 2-5 Primary diffuser and liquid gas separator.

PAGE 30

20 1.3.8 Fan Assembly A single width, single inlet (SWSI) centrifugal fan (TCF/Aerovent Model No. 16 BIA-SW-ARR: 10 CL2 CW THD) is used. The fan is arranged in a top horizontal discharge configuration in which the inlet is perpendicular to the direction of airflow. When viewed along the direction of flow, the air intake is from the left. 1.3.9 Motor and Motor Controller The fan is belt driven by a 7.5 HP, ODP, 3600 RPM, 208-230 VAC, 3 60 Hz motor (Toshiba Model No. BY752LF2UMH04). A transistor inverter type, variable speed motor control (Toshiba Model No. VFS92055PL-WN) controls the fan shaft RPM. The controller is arranged for 208-230 VAC, 3 60 Hz input electric service. A remote control station, located adjacent to the test section, allows easy regulation of the test section speed. A NEMA rate fusible disconnect ensures the motor and controller are safe. The disconnect is mounted in a NEMA Type 4 enclosure. 1.3.10 Supporting Frame The wind tunnel is supported by structural steel tubing frames that are welded together. The frames are etched, prime coated and spray finished with acrylic enamel. Leveling pads and swivel casters are fitted for height adjustment and transportation. To minimize conduction of vibration from room and fan/pump, rubber-in-shear mounts are provided (Figure 2-6).

PAGE 31

21 Figure 2-6 Rubber-in-shear mounts to minimize vibrations. The next chapter will describe measurement velocity profiles for both the air and water tunnels. Also, measurement of vibration characteristics of the test section as a result of the fan and pump is described.

PAGE 32

22 CHAPTER 3 CHARACTERIZATION OF FACILITY To characterize the wind and water tunnel test sections, the mean flow quality of both the sections and the influence of vibration of the fan and pump on the test section vibration are considered here. The mean velocity profiles of both the test sections are measured separately. For the velocity measurements, a pitot-static probe is used in conjunction with a differential pressure transducer. Vibration measurements are carried out by placing accelerometers at tentative vibration sources, namely the fan and the pump, and the mid region of the test section. The velocity profile measurement for the air tunnel is described below. Velocity Profiles of the Air Tunnel 1.3.11 Experimental Set-up The experimental set-up for measuring the velocity profile of the air stream is shown in Figure 3-1. The pitot-static probe is inserted from the slot in the ceiling of the test section at a distance of 19 in. from the start of the test section. The mounting of the pitotstatic probe and the coordinate system defined are shown in Figure 3-2. To avoid the air water interface, the tunnels are isolated using a plexiglas cover. Along with the pitotstatic probe, two pressure transducers, Validyne P855A and a Heise HSQ-1 (Range: 0 – 15 in. H2O), are used to measure the dynamic head of the flow. A T-junction is used to split the pressure lines from the pitot-static probes to connect to the Heise and the Validyne transducers. The voltage output from the Validyne P855A, in the range from – 5 to +5 V DC, is read using a HP 34970A integrating voltmeter. 22

PAGE 33

23 Figure 3-1 Experimental set-up used to measure velocity profile in air tunnel. FLOOR CEILING 3.5 in. 0.125 in. 6.0 in. 18.0 in. x y Splitter plate 19.0 in. Pitot static probe Figure 3-2 Schematic of the pitot-static probe mounting in the test section.

PAGE 34

24 Additional apparatus included a micron resolution 3D traverse system for precisely positioning the pitot-static probe and an NI PCI6024E series DAQ card with programmable voltage output for controlling the speed of the air tunnel. The output voltage from the DAQ card transforms to a frequency of operation for the motor that operates the fan for the air stream. A LabVIEW program is used to integrate 1) data acquisition from the HP 34970A DAQ unit over a GPIB interface, 2) data acquisition from the Heise pressure gauge with a serial port connection and 3) control the NI PCI6024E output voltage, to set the air speed. The operating instructions for this LabVIEW program as well as the tunnel operating instructions are explained in Appendix A. 1.3.12 Calibration of Air Tunnel Velocity and Air Pressure Transducer The pitot-static probe is fixed at the mid-height of the tunnel test section. The block diagram for calibration is given shown in Figure 3-3. In an automated process, the input voltage, va, to the motor is increased from 0 10 V DC in steps of 0.2 V. For each va, 25 readings are taken from the Heise gauge and the HP 34970A DAQ unit over a period of approximately 10 seconds. These 25 readings are averaged to obtain a single reading from the Heise Gauge and HP 34970A DAQ unit for that particular va. The result recorded achieves calibration of the wind tunnel speed with the DAQ voltage input and is used for all subsequent measurements. This is shown in Figure 3-4. A look-up table for va versus the motor frequency is recorded to facilitate manual setting of desired speed based on frequency. The dependence is linear except at very low input voltage. From this, the threshold value of the input is found to be approximately 0.6 V.

PAGE 35

25 DAQ computer w/ NI 6024E DAQ Pitot Static Probe P855A Pressure Transducer HP 34970A DAQ Unit Heise Gague HSQ-1 Air Tunnel Figure 3-3 Block diagram for calibration and velocity profile measurement for the air tunnel. 0 2 4 6 8 10 0 10 20 30 40 50 60 70 DAQ voltage (Volts)Motor frequency (Hz) 0 2 4 6 8 10 0 10 20 30 40 50 60 70 Velocity (m/s) Figure 3-4 Motor frequency and velocity of air tunnel plotted as a function of DAQ input voltage. Data from the HP 34970A DAQ unit (output voltage from the Validyne P855A) is used to calibrate the P855A with respect to the Heise gauge. 1.3.13 Velocity Profiles The block diagram for the velocity profile measurements is shown in Figure 3-3. For the velocity profile measurement, the pitot-static probe is traversed from the bottom of the tunnel to the top of the tunnel in steps of 2 mm while keeping va constant. Then, va is steadily increased from 0 10 V DC in steps of 0.2 V. An average of 25 readings from

PAGE 36

26 the Heise gauge are recorded for each va and each traverse position. The results are plotted as velocity profiles as shown in Figure 3-5. The velocity profiles do not reach zero at the floor because of the finite diameter ( 1 8 in.) of the pitot-static probe. 0 1 2 3 4 5 6 0 10 20 30 40 50 60 70 Tunnel profile (inches)Velocity (m/s)Floor Ceiling Figure 3-5 Velocity profiles (with uncertainity estimates) of the air tunnel for velocities increasing from 6 m/s to 66 m/s in ten steps (corresponding to motor input voltage increasing from 1 V to 10 V in steps 1 V). Table 3-1 summarizes the results of the velocity profile measurements for the air tunnel. The mean, maximum percent absolute deviation and standard deviations are tabulated outside the local boundary layer. At a particular operating condition, the maximum absolute deviation is calculated as the largest absolute magnitude of the difference between the velocity measurements and the mean value divided by the mean value. This deviation is found to reduce as the mean speed increased while the standard deviation values increased progressively.

PAGE 37

27 Table 3-1 Air tunnel velocity characteristics Mean velocity (m/s) % Maximum absolute deviation % Standard deviation (with respect to mean) 6.1 2.17 0.98 12.4 0.76 0.32 19.0 0.70 0.31 25.7 1.19 0.47 32.4 0.82 0.34 38.9 0.75 0.39 45.4 0.80 0.42 51.7 0.75 0.35 58.0 0.91 0.33 64.2 0.70 0.28 Velocity Profiles of the Water Tunnel 1.3.14 Experimental Set-up The experimental set-up used here is similar to that of the air measurements except that a free air-water surface is maintained in this case. The pitot-static probe is inserted from the slot in the ceiling of the test section at a distance of 10 in. from the start of the test section. The tubes from the probe are connected to the transducer after flooding all parts with water. Visual inspection insured no bubbles were present. The pitot-static probe is connected to Validyne P55D pressure transducer that has a full-scale range 2.5 in. of H2O. The output of this transducer is connected to the HP 34970A DAQ unit. The 3D traverse with servomotors is used to position the probe. An NI PCI6024E series data acquisition card with programmable voltage output is used for controlling the frequency of the water pump, thereby controlling the speed of the water tunnel. An integrated

PAGE 38

28 LabVIEW program is used to 1) control the water tunnel speed, 2) collect data from the HP Unit using GPIB interface and 3) control the traverse. The operating instructions for this LabVIEW program is given in Appendix A. 1.3.15 Calibration of Water Tunnel Velocity and Water Pressure Transducer The pitot-static probe remains fixed at the center of the air tunnel test section. Access to the air water interface is blocked by placing a plexiglas cover over the water tunnel. The block diagram for calibration is shown in Figure 3-6. Input voltage va to the motor is increased in small intervals from 0 until a point where the dynamic head of the air stream is a little less than 2.5 in. of H2O. For each va, 25 readings are taken from Heise Gauge and also from HP 34970A DAQ unit over a period of approximately 10 seconds. Each of these 25 readings is averaged to get a single reading from the Heise Gauge and HP 34970A DAQ unit for that particular va. The result recorded achieves the calibration of the P55D with respect to the Heise gauge. To calibrate the water tunnel velocity with respect to the motor frequency, the pitotstatic probe is positioned at the center of the water tunnel. The input voltage, vw, to the pump motor is increased from 0-10 V DC in steps of 0.2 V. For each vw, 25 readings are taken from the P55D using the HP 34970A DAQ unit over a period of approximately 10 seconds. Each of these 25 readings is averaged to get a single reading for a particular vw. The result recorded achieves calibration of the water tunnel speed with the DAQ voltage input and is used for all measurements in future. This is shown in Figure 3-7. A look-up table for vw versus the motor frequency is recorded simultaneously to facilitate manual setting of desired speed based on frequency. The dependence is linear except at very low

PAGE 39

29 input voltage. From this, the threshold value of the input is found to be approximately 0.6 V. DAQ computer w/ NI 6024E DAQ Pitot Static Probe P55D Pressure Transducer HP 34970A DAQ unit Heise Gague Water Tunnel Air Tunnel Calibration Figure 3-6 Block diagram for calibration and velocity measurement for the water tunnel. 0 2 4 6 8 0 10 20 30 40 50 60 DAQ voltage (Volts)Motor frequency (Hz) 0 2 4 6 8 0 0.2 0.4 0.6 0.8 Velocity (m/s) Figure 3-7 Motor frequency of water tunnel plotted as a function of DAQ input voltage. 1.3.16 Velocity Profiles For the velocity profile measurement of the water tunnel, the pitot-static probe is traversed from the bottom of the water tunnel until just below the air water interface in steps of 2 mm. The voltage vw is steadily increased from 1-8 V DC in steps of 1 V. An

PAGE 40

30 average of 25 readings from the water pressure transducer using the HP 34970A DAQ unit are recorded for each vw and each traverse positions near the floor. The resulting velocity profiles are shown in the plot in Figure 3-8. Large uncertainties are observed, particularly at high speeds and probe positions, because of vibration experienced by the probe due to the air-water interface. For both air and water streams, more accurate measurements like particle image velocimetry and Laser Doppler velocimetry are required to determine the basic flow pattern. The resulting velocity profiles can be used to develop analytical models of the air-water interface. 0 1 2 3 4 5 6 0 0.2 0.4 0.6 0.8 1 Velocity (m/s)Tunnel profile (inches) Floor Interface Figure 3-8 Velocity profiles (with uncertainty estimates) of water tunnel for velocity increasing from 0.15 m/s to 0.86 m/s in seven steps (corresponding to motor input voltage increasing from 1 V to 8 V in steps of 1 V). Table 3-2 summarizes the results of the velocity profile measurements for the water tunnel. The mean, maximum absolute deviation and standard deviations are tabulated as with the air measurements. The maximum absolute deviation is found to fluctuate

PAGE 41

31 without any pattern the standard deviation values steadily increased as the mean velocity increased. Table 3-2 Water tunnel velocity characteristics Mean velocity (m/s) % Maximum absolute deviation % Standard deviation (with respect to mean 0.16 1.02 0.56 0.26 1.43 0.54 0.38 4.30 1.18 0.50 2.19 0.86 0.61 1.36 0.51 0.74 2.30 1.22 0.82 2.21 0.68 0.85 0.90 0.49 Vibration Measurements of the Test Section The vibrations induced by the fan and pump, especially at high operating speeds, may vibrate the test section to a reasonable degree. Hence, the vibration influence of the fan and pump on the test section may become critical to the flow quality of both the air and water stream in the test sections. Therefore, vibration measurements at select operating conditions are performed. 1.3.17 Experimental Set-up Three tri-axial accelerometers (Piezotronic Model 356A16) are used to measure the vibration levels. Two accelerometers are mounted close to the source of the vibration, i.e., at the fan and the pump as shown in Figure 3-9. A third accelerometer is placed approximately at the midpoint of the test section as shown in Figure 3-10. The accelerometers are mounted so as to measure accelerations in a coordinate system that is

PAGE 42

32 irrespective of any accelerometer placement is defined as shown in Figure 3-9 and Figure 3-10. All accelerometers are attached using wax. (a) blower (b) pump Figure 3-9 Accelerometers mounted at possible vibration sources, i.e. at the (a) blower and the (b) pump respectively. Figure 3-10 Accelerometer mounted approximately at the midpoint of the test section to detect vibration levels. Flow direction is along X. A signal conditioner is used prior to acquiring the data from the accelerometers. A HP E1433 16-channel 52 kHz digitizer is used for acquiring the data. Each channel has a built-in programmable anti-aliasing filter whose cutoff span, fc, is determined by the sampling frequency, fs, as 2.56csff (3.1)

PAGE 43

33 The data is recorded for the operating conditions listed in Table 3-3. The frequency values in Table 3-1 correspond to the inverter frequency that is used to set the speed of the fan and pump impeller. Table 3-3 Test matrix for vibration measurement Air (Hz|ms-1) Water (Hz|ms-1) 00|00.0 30|32.4 60|64.5 00|0.0 30|0.5 60|0.9 For all cases, the acceleration vector (i.e., all three orthogonal components) is measured. The signal is sampled at 2048 samples per second. For a sampling rate of 2048, a usable span of 800 Hz results. A total record length of 393216 points is collected. The record length is broken into 192 blocks of 2048 samples, resulting in a frequency bin of 1 Hz, to perform analysis. A hanning window is applied to the data and an overlap ratio of 75% is used. The estimate of random error with 192 averages is 0.072. 1.3.18 Theoretical Model The system is mathematically modeled as a two-input single-output system as shown in Figure 3-11 [Bendat and Piersol 2000]. In Figure 3-11, x1(t) corresponds to the accelerometer data from the fan, x2(t) corresponds to the accelerometer data from the pump, n(t) is uncorrelated noise (with the inputs x1(t) and x2(t) ) and y(t) is the vibration measured at the test section. Since the tri-axial accelerometer gives accelerations in three directions, x1, x2, and y can be only one of the three components. In order to choose the

PAGE 44

34 component, the power spectra of the three components for all three accelerometers are plotted, as shown in Figure 3-12 below. H1(f) H2(f) x1(t) x1(t) n(t) y(t) Figure 3-11 Two-input single output model for vibration analysis. 0 100 200 300 400 500 600 700 800 10-10 10-5 100 FAN z axis x axis y axis 0 100 200 300 400 500 600 700 800 10-10 10-8 10-6 10-4 10-2 PUMP z axis x axis y axis 0 100 200 300 400 500 600 700 800 10-10 10-8 10-6 10-4 10-2 Frequency (Hz) TEST SECTION z axis x axis y axis Figure 3-12 Power spectra of three components for all three accelerometers.

PAGE 45

35 Based on Figure 3-12, the x component of accelerometer mounted near the fan is chosen as x1, the y component of accelerometer mounted near the pump is chosen as x2, and the x component of accelerometer mounted on the test section is chosen as y The system shown in Figure 3-11 can be expressed in frequency domain as 1122YHXHXN (3.2) where X1, X2, N and Y are the Fourier transforms of x1, x2, n and y respectively. The relevant power and cross spectra are given by 1212 11 22 *2 () 2 () 2 () 2 ()y y yyGfEXX T GfEXY T GfEXY T GfEYY T (3.3) where denotes the complex conjugate. The ordinary coherence functions are calculated as follows: 2 12 2 12 1122 2 1 2 1 11 2 2 2 2 22() () ()y y y y y y y yG f GG G f GG G f GG (3.4) Assuming 2 12() f (0,1), the noise spectrum Gnn is computed as 22 ** 11112122121222nnyy nnyyvvGGHGHHGHHGHG GGG (3.5)

PAGE 46

36 where the frequency response functions H1 and H2 are calculated according to the formula given below: 221122 1 2 112212 112211 2 2 112212 y y y yGGGG H GGG GGGG H GGG (3.6) The cross terms involving H1 and H2,, i.e. ** 12122121 and H HGHHG, arise because of the interaction between x1(t) and x2(t) The multiple coherence function, that represents the correlated output power in the test section acceleration due to the input fan and pump acceleration, is given by 2 :()vv yx y yG f G (3.7) 1.3.19 Results and Conclusions Results of the vibration data for the case when the air speed is 64.5 m/s and water speed is 0.5 m/s are presented. Results for all other test cases, tabulated in Table 3-1, are presented in Appendix B. The ordinary coherent output spectra, 2 1 y yyG and 2 2 y yyG are shown in Figure 3-13. Also, the noise contributions Gnn and the multiple coherent output spectrum (i.e., the combined vibration contributions of the fan and pump) Gvv are compared to the output spectrum Gyy in Figure 3-13. For comparison, the measured noise floor spectrum, obtained when the blower and pump are off, is also plotted.

PAGE 47

37 The ordinary coherence functions between the inputs, 2 12 and between each input and the output, 2 1 y and 2 2 y are shown in Figure 3-14. Also, the multiple coherence function, 2 : y x is shown in Figure 3-14. The results indicate a large content of the output spectrum is a part of the "noise" Gnn that is not modeled. However, the multiple coherence function (and other ordinary coherence functions) is close to 1 at certain frequencies indicating that contribution at these frequencies are largely due to the stated inputs x1 and x2, i.e. the z axis component of the vibration measured by the accelerometers near the fan and pump, respectively.

PAGE 48

38 0 100 200 300 400 500 600 700 800 10-10 10-5 2 1y Gyy 0 100 200 300 400 500 600 700 800 10-10 10-5 2 2y Gyy 0 100 200 300 400 500 600 700 800 10-10 10-5 0 100 200 300 400 500 600 700 800 10-10 10-5 Frequency (Hz) GnnG noise floor GyyGvvGnn Figure 3-13 Top two plots show the ordinary coherent output spectrum due to the inputs. The bottom two plots compare the noise spectrum with the output spectrum, the multiple coherent output spectrum and the noise floor.

PAGE 49

39 0 100 200 300 400 500 600 700 800 0 0.2 0.4 0.6 0.8 1 2 12 0 100 200 300 400 500 600 700 800 0 0.2 0.4 0.6 0.8 1 2 1y 0 100 200 300 400 500 600 700 800 0 0.2 0.4 0.6 0.8 1 2 2y 0 100 200 300 400 500 600 700 800 0 0.2 0.4 0.6 0.8 1 2 xyFrequency (Hz) Figure 3-14 The topmost plot shows 2 12 the coherence between the inputs. The next two plots are 2 1 y and 2 2 y the ordinary coherence functions between each input and the output. The bottommost plot is 2 : y x the multiple coherence function. To distinguish the effect of x1 from x2 on the output, the characteristics of the blower and pump are studied. For the blower and the pump, the blade passage frequencies of the

PAGE 50

40 fan and the pump impeller are identified in a separate experiment using a BK condenser microphone. The blade passage frequencies of the fan and pump impeller are determined for the conditions listed in Table 3-1. To determine the blade passage frequency of the fan, the pump is switched off and acoustic pressure measurements are obtained very close to the entrance of the blower with the microphone. A similar exercise is repeated for the pump impeller. The specifications of the microphone are given in below. The results are compared to Gyy in Figure 3-15. A strong tone at 364 Hz is noticed in the spectrum of the microphone data near the blower. A second tone of lesser intensity is noticed at 728 Hz in this spectrum. The second tone is inferred as the harmonic of the 364 Hz tone. In addition, a range of smaller tones is noticed at 40 Hz, 81 Hz, 121 Hz, 162 Hz and 202 Hz. This is explained as follows. The fan in the blower is known to have nine blades. The blade passage frequency is calculated as the number of blades times the frequency of revolution of the fan. The result of dividing the frequency of the first tone, i.e. 364, by the number of blades is very close to 40 Hz. Therefore, there is a high probability that the first peak of the range is a result of eccentricity in the mounting of the fan because it is likely to have the effect of producing tones at the frequency of revolution. The subsequent peaks at multiples of 40 Hz are most likely to be the harmonics of the tone at 40 Hz. The output spectrum Gyy has a strong tone at 364 Hz and 728 Hz, which is mostly due to the strong vibration exhibited by the blower at the blade passage frequency and its first harmonic. Also, discernible peaks are noticed at 40 Hz, 81 Hz and 202 Hz that arise

PAGE 51

41 mostly as an effect of eccentricity in the mounting of the fan in the blower section. In addition, several peaks are noticed spaced approximately 40 Hz apart after 364 Hz. 0 100 200 300 400 500 600 700 800 10-14 10-12 10-10 10-8 10-6 10-4 10-2 Frequency (Hz) GyyFAN MIC SPECTRUM FAN MIC NOISE SPECTRUM 0 100 200 300 400 500 600 700 800 10-14 10-12 10-10 10-8 10-6 10-4 10-2 Frequency (Hz) GyyPUMP MIC SPECTRUM PUMP MIC NOISE SPECTRUM Figure 3-15 Comparison of output spectrum with BK microphone measurements. In the top plot, the microphone is placed very close to the blower, with the pump off. In the bottom plot, the microphone is placed very close to the pump, with the blower off. For the case when the air speed is 60 m/s and water speed is 0.5 m/s, the comparison of the microphone spectrum near the pump with the output vibration spectrum Gyy is

PAGE 52

42 inconclusive. This is because vibration levels induced by the pump are very small compared to the fan. The ordinary coherence function (2 12 ) between the inputs, plotted in Figure 3-14, shows that the pump experiences significant vibration from the fan. To isolate the influence of the pump vibrations on the on the test section, the vibration data for the case when the fan is switched off is analyzed. The water tunnel speed is set at approximately 0.5 m/s. Along with the vibration measurements, BK microphone measurements are made. The results are shown in Figure 3-16. 0 100 200 300 400 500 600 700 800 10-14 10-12 10-10 10-8 10-6 10-4 10-2 Frequency (Hz) GyyPUMP MIC SPECTRUM PUMP MIC NOISE SPECTRUM Figure 3-16 Comparison of the output spectrum with microphone measurements for the case when the fan is off and the water tunnel speed is 0.5 m/s. The output spectrum is almost three orders of magnitude less than that of the previous case analyzed. The results show that harmonics as well as sub harmonics are present in both the output spectrum and the microphone data.

PAGE 53

43 In conclusion, the fan vibrations are much greater than the pump. The test section vibration after removing the fan and pump contributions is still much greater the measured than noise floor, indicating other significant inputs (e.g., the other vibration fan and pump components).

PAGE 54

44 CHAPTER 4 INTERFACE EDUCTION USING LIGHT SHEET This chapter describes a simple method used to characterize the air water interface of the facility under different combinations of air and water speeds. The results are helpful in determining the amount of fluctuations in the air-water surface at a particular condition. A measure of both root mean squared amplitude and slope are obtained using a high-resolution camera in conjunction with a pulsed laser light sheet. A simple edgedetection algorithm is employed to determine the location of the air-water interface. The aim is to educe quantitative information of the air-water interface. Experimental Set-up The experimental set-up consists of a pulsed Nd:YAG laser and a digital camera (TSI model PIVCAM), as shown in Figure 4-1. The Nd:YAG and associated optics generate a thin light sheet (~1mm thick) required to illuminate the air-water interface. The camera resolution is 1016 by 1000 pixels and acquires image pairs that are synchronized with the laser. To enhance the scattering at the air-water interface, a fluorescent dye Eosine Y is used. Some of the critical parameters used with the laser and camera are listed in Table 4-1. A span of little less than 4.5 in. is covered (corresponding to the range of 5.5 in.-10 in. from the splitter plate), yielding a spatial resolution of 209 pixels/in. and 223 pixels/in. in horizontal and vertical world coordinates, respectively. As shown in the camera is Figure 4-1, the camera is positioned to obtain a perspective view of the air-water surface. A perspective view avoids an overlap between the desired 44

PAGE 55

45 mid-plane edge and the unwanted edge at the sidewall. This simplifies the air-water interface eduction algorithm. Figure 4-1 Experimental set-up with Nd:YAG LASER and PIVCAM Table 4-1 Critical parameters of the laser and camera Pulse separation ( t) 500 s Camera shutter open time 20 s Pulse rep rate (PRR) of laser 15 Hz (max) Pixel array size 1016 1000 1.3.20 Procedure A sequence of images at the conditions shown is acquired for the conditions listed in Table 4-2. The numbers in the table correspond to the speed of the air/water streams. A total of 200 images are taken at the pulse repetition rate (PRR) specified in Table 4-1.

PAGE 56

46 Table 4-2 Test matrix for interface eduction Air (Hz|ms-1) Water (Hz|ms-1) 0|0.0 3|3.1 5|5.2 Pump off 0|0.0 0.1 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 An edge detection algorithm is applied to detect the air-water interface from the images. The edge detection algorithm used is based on the sobel method [Gonzalez and Woods 2001]. An example for the case with an air speed 5.2 m/s and a water speed 0.4 m/s is shown in Figure 4-2. In all images, the flow direction is from left to right.

PAGE 57

47 200 400 600 800 1000 200 400 600 800 1000 X pixels Y pixels 200 400 600 800 1000 200 400 600 800 1000 Y pixels X pixels Figure 4-2 Sample original image (left) and corresponding output from edge detection algorithm (right). The flow direction is from left to right as indicated. Since the edge detection algorithm detects all edges, several extraneous edges can be identified as shown in Figure 4-2. The fact that sufficient scattering occurs at the airwater interface ensures that the interface is always a nearly continuous edge. This fact is exploited to prune the image of the unwanted edges or features. 1.3.21 Interface Detection The key idea is to obtain the interface image as a binary image (i.e., the interface pixel has a value of 1 and 0 otherwise). The output of the edge detection algorithm can be pruned to provide the interface image shown in Figure 4-3. Since the interface is detected as a clear continuous edge, the occurrence of the first pixel value of 1 found in a column search of the edge image is assumed to be the interface.

PAGE 58

48 0 200 400 600 800 1000 300 350 400 450 500 550 Y pixels X pixels Mid-plane interface Side-wall edge 200 400 600 800 1000 200 400 600 800 1000 X pixels Y pixels Mid-plane interface Figure 4-3 A sample edge image and the corresponding interface image as a result of pruning. The interface images of a whole sequence of 200 are then averaged, thereby obtaining an average interface for a particular tunnel condition. Also, the standard deviation value of the fluctuations of the interface location is calculated as 2 10001000 11 10001000 11 jj jj i jj jjIKKIK II (4.1) where j I is the a column vector of Y pixel values at jth X pixel, [1,2,...,1016] K denotes element by element multiplication and denotes rounding off to the nearest integer value. The advantage of computing the standard deviation value is two fold. First, it gives a measure of the fluctuations of the air-water interface. Second, it is essential to identify outlier points. An average interface image and the corresponding standard deviations (from mean at an average interface image pixel coordinate) are shown in Figure 4-4.

PAGE 59

49 200 400 600 800 1000 200 400 600 800 1000 X pixels Y pixels 0 200 400 600 800 1000 0 5 10 15 20 25 30 Standard deviation valuesX pixels Figure 4-4 Average interface and standard deviation of air-water interface. Number of statistics: 200 1.3.22 Outlier Rejection The interface detection algorithm reveals outliers as shown in Figure 4-5. The presence of outlier points in the edge image results in high standard deviation values at certain X pixels as shown in Figure 4-4. The outlier points and the interface pixel distribution for a column of sum of sequence of interface images are shown in Figure 4-5. 200 400 600 800 1000 200 400 600 800 1000 Outlier points Y pixels X pixels 0 200 400 600 800 1000 0 5 10 15 20 25 30 Y pixelsInterface points of sequenceOutlier point Figure 4-5 Superposition of sequence of edge images before outlier rejection and a column (at 188th X pixel) of sum of sequence of interface images with an outlier point.

PAGE 60

50 The outlier criteria used here is based on comparing standard deviation. First, the mean and standard deviation of the data set are calculated for each column in an image. Then, all data points that lie outside the interval 3,3 are disregarded according to Modified Tau Thompson criterion [Holman 2000]. Applying this to the sum of sequence of interface images, the result, shown in Figure 4-6, is obtained. 200 400 600 800 1000 200 400 600 800 1000 X pixels Y pixels 0 200 400 600 800 1000 0 5 10 15 20 25 30 Y pixelsInterface points of sequence Figure 4-6 Sum of sequence of edge images after outlier rejection and a column (at 188th X pixel) of sum of sequence of interface images without any outlier points. After the outliers are eliminated, the interface image and standard deviation values are calculated again. The outlier criterion does not significantly affect the mean of the data set; hence the average interface image is almost unchanged (Figure 4-7). Analysis of the two images in Figure 4-7 shows that the difference between the interfaces is not more than one pixel.

PAGE 61

51 200 400 600 800 1000 200 400 600 800 1000 Y pixels X pixels 200 400 600 800 1000 200 400 600 800 1000 X pixels Y pixels Figure 4-7 Comparison of interface images without and with outlier rejection. The average images are shown before outlier rejection (left) and after outlier rejection. However, the application of outlier criterion has significant effect on the standard deviation of the interface image. The standard deviation values before and after outlier rejection are shown in Figure 4-8. 0 200 400 600 800 1000 0 5 10 15 20 25 30 Standard deviation valuesX pixels 0 200 400 600 800 1000 0 5 10 15 20 25 30 X pixels Standard deviation values Figure 4-8 Comparison of standard deviation values before and after outlier rejection. 1.3.23 Transformation and Calibration Since the images are taken with a camera, the original spatial signal (image) undergoes a perspective transformation. Assuming the camera is a point source, a

PAGE 62

52 reverse projection transformation is required to map the image to world coordinates so that the interface location is determined in physical dimensions. For this purpose, a calibration image is obtained with the camera. In the calibration image, the world coordinates of specific points in the image are known. In this case, the calibration image is a series of squares whose vertices are known in world coordinates. The lines of the square that are captured by the camera are at least 5 pixels thick. Projection transformation (camera view) of these squares will transform parallel lines into lines that intersect at some point within or outside the image. The "n" number of intersection points defines the projection transformation as an "n" point projection. Since the image is of a plane as a pixel map in this case, the projection is a two-point projection. The camera view (transformed) and the reverse projection transformation of the calibration image are shown in Figure 4-9. The transformation is achieved by means of trial and error by ensuring the quadrilaterals are mapped to rectangles whose vertices have a pixel difference of less than 1. 300 400 500 600 700 800 150 250 350 X pixels Y pixels 300 400 500 600 700 800 150 250 350 X pixels Y pixels Figure 4-9 Projection transformation and reverse projection transformation of the calibration image. The following transformation matrix achieves the desired result:

PAGE 63

53 0.93220.02730.0273 0.04700.95170.0405 0.04700.00781.000 T (4.2) 000.0470.0078 100.910.02 0110 1111.001 xyxy T (4.3) where (,) x y are the world coordinates and (,) x y are image coordinates. Calibration of the reverse projection transformation image enables measuring the interface in physical dimensions (inches) instead of pixels. Knowing that the squares in the calibration image are of length 1 in., the following results: World X axis: 209 pixels = 1 in. World Y axis: 223 pixels = 1 in. (4.4) Now, applying the reverse transformation to the average interface image, we get the average interface in the world coordinate system as shown in Figure 4-10. 200 400 600 800 1000 200 400 600 800 1000 X pixels Y pixels 5 6 7 8 9 10 -2 -1 0 1 2 X inchesY inches Figure 4-10 The average interface image and the corresponding reverse transformation to world coordinates.

PAGE 64

54 1.3.24 Interface Slope Calculation The first step in calculating the slope of the interface is fitting a polynomial of an appropriate order. A visual inspection of the interface image (Figure 4-11) shows that there are four extremums. Therefore, a polynomial fit of 5th order is chosen as shown in Figure 4-11. This order is chosen to ensure minimal error in the polynomial fit while maintaining a smooth slope. The value of maximum error for possible polynomial fits is tabulated in Table 4-3. Table 4-3 The maximum error between the data and polynomial fit. Polynomial fit orderMaximum error (pixels) 3 5.1 4 3.8 5 1.4 6 1.5 7 1.8 8 1.5 5 6 7 8 9 10 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 X inchesY inches 5 6 7 8 9 10 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 Interface polyfit X inchesY inches Figure 4-11 Interface image in world coordinates and corresponding polynomial fit. With the polynomial fit known, the slope of the interface can be calculated by differentiating the expression. The corresponding slope is also shown in Figure 4-12.

PAGE 65

55 Comparison of every case with the baseline case of zero velocity of air and water tunnel will give the relative interface displacements. 5 6 7 8 9 10 -2 -1 0 1 2 Interface polyfit X inchesY inches 5 6 7 8 9 10 -5 -2.5 0 2.5 5 x 10-3 X inchesSlope Figure 4-12 Polynomial fit, of order 5, of average interface image (left) and corresponding slope of interface (right). The next chapter will summarize with conclusions and also list the possible future endeavors related to the facility.

PAGE 66

56 CHAPTER 5 CONCLUSIONS AND FUTURE WORK The University of Florida air-water shear-layer facility is unique for testing algorithms and characterizing MEMS sensors related to supercavitating vehicles. The facility allows for sensor characterization and algorithm testing by emulating supercavitating conditions satisfactorily. Velocity profiles measured for the air stream and the water stream are found to be uniform. The mean flow quality is very good in both the sections. Future work can be done for more accurate measurements of the velocity profile, variations in speed as well as angular using more precise measurement tools such as particle image velocimetry and LASER Doppler velocimetry. Also, influence of fan and pump vibration on the fluctuating flow quantities has to be assessed. Operation of the air tunnel at high speeds, with free surface condition, entrains significant amount of water from the water tunnel thereby rapidly reducing the water level in the water tunnel. As a remedy to this problem, a feedback-based system should be employed to maintain desired water levels in the water tunnel. Such a feedback system would comprise of a variable throughput pump, a water level sensor with very low reaction time and a feedback circuit. 56

PAGE 67

57 APPENDIX TUNNEL OPERATION PROCEDURE 1. Set Air speed control to "auto." 2. Leave the channel as channel "0." 3. Choose the voltage increments and ceiling to be used for the experiment, with a maximum ceiling of 10 Volts. 4. Set the file path to which data will be written. 5. Choose the appropriate IP address for the traverse to be used. 57

PAGE 68

58 6. On the S7 Operator Interface panel on the side of the air tunnel, press the "MON" button. 7. Hit "ENT" to enter the "Basic Parameters Group." 8. Press the up arrow four times, to access the "Frequency select" menu. Press "ENT." 9. Press the up arrow to bring the displayed setting to "Terminal." Press "ENT." This brings the tunnel under DAQ control. 10. Press "MON" twice to return to the main display. 11. Press the "RUN" button on the panel. 12. Run the VI. 13. When the experiment is complete, press the "STOP" button on the panel.

PAGE 69

59 1. Set Air speed control to "auto." 2. Set the Channel to channel "1," to enable water speed control instead of air. 3. Choose the voltage increments and ceiling to be used for the experiment, with a maximum ceiling of 8 Volts. 4. Set the file path to which data will be written. 5. Choose the appropriate IP address for the traverse to be used. 6. On the S7 Operator Interface panel on the side of the water tunnel, press the "MON" button. 7. Hit "ENT" to enter the "Basic Parameters Group." 8. Press the up arrow four times, to access the "Frequency select" menu. Press "ENT." 9. Press the up arrow to bring the displayed setting to "Terminal." Press "ENT." This brings the tunnel under DAQ control. 10. Press "MON" twice to return to the main display. 11. Press the "RUN" button on the panel. 12. Run the VI. 13. When the experiment is complete, press the "STOP" button on the panel.

PAGE 70

60 REFERENCES Banazynski, K. A. (2000). Air/Water Shear Layer Facility Manual, Engineering Laboratory Design Inc., St. Louis, MN. Bendat, J. S., Piersol, A. G. (2000). Random Data: Analysis and Measurement procedures, Wiley-Interscience, New York. Billet, M., Tip vortex cavitation, http://www.arl.psu.edu/areas/cavitation/cavitation.html Accessed on 06/10/2002. Chandrasekaran, V., Chow, E. M., Kenny, T. W., Nishida, T., Sheplak, M. (2001a). “Through Wafer Electrical Interconnects for MEMS Sensor,” Proceedings of the American Society of Mechanical Engineers. New York. NY. pp 232-245 Chandrasekaran, V., Li, X., Nishida, T., Cattafesta, L. N., Li, J., Sheplak, M. (2001b). “Thermoelastically Actuated MEMS Ultrasonic Transducer,” Presented at 142nd Meeting of the Acoustical Society of America. Ft. Lauderdale, FL. Chandrasekaran, V., Chow, E. M., Kenny, T. W., Nishida, T., Cattafesta, L. N., Sankar B.V., Sheplak, M. (2002). “Thermoelastically Actuated Acoustic Proximity Sensor With Integrated Through-Wafer Interconnects,” Presented at Solid-State Sensor and Actuator Workshop. Hilton Head. SC. Gonzalez, R. C. and Woods, R. E. (2001). Digital Image Processing, 2nd edition, Addison-Wesley, New York. Holman, J. P. (2000). Experimental Methods for Engineers, McGraw Hill Higher Education, New York. Kulkarni, S. S., Pratap, R. (1999). "Studies on the Dynamics of a Supercavitating Projectile," Applied Mathematical Modeling 24: 113-129. Kunz, R. F., Lindau, J. W., Billet, M. L., Stinebring, D. R. (2000). “Multiphase CFD Modelling of Developed and Supercavitating Flows,” Applied Research Laboratory, Pennsylvania State University. Li, X., Lasson, E., Sheplak, M., Li, J. (2002a). “Phase-Shift-Based Time Delay Estimators for Proximity Acoustic Sensors,” IEEE Journal of Oceanic Engineering. 27: No. 1. 47-56. 60

PAGE 71

61 Li, X., Wu, R., Sheplak, M., Li, J. (2002b). “Multifrequency CW-Based Time-Delay Estimation for Proximity Ultrasonic Sensors,” To appear in IEE Proceedings F: Radar, Sonar, and Navigation. Li, X., Wu, R., Rasmi, S., Li, J., Cattafesta, L., Sheplak, M. (2002c). “An Acoustic Proximity Ranging System for Monitoring the Cavity Thickness,” Submitted to IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control. Rightley, P. M, Lasheras, J. C. (2000). “Bubble Dispersion and Interphase Coupling in a Free-Shear Flow,” Journal of Fluid Mechanics 412: 21-59. Sheplak, M., Cattafesta, L. N., Shyy, W., Kurdila, A. J., Nishida, T. (1999). Advanced Technology Development for the Control of High-Speed Supercavitating Vehicles, Interdisciplinary Microsystems Group, Mechanical and Aerospace Department, University of Florida. Stinebring, D. R., Cook, R. R., Dzielski, J. E., Kimerer, N. B., Kunz, R. F., Miller, T. F. (2000). High-Speed Supercavitating Vehicles, Applied Research Laboratory, Pennsylvania State University. Stinebring, D. R., Billet, M. L., Lindau, J. M, Kunz, R. F. (2001). Developed CavitationCavity Dynamics, Applied Research Laboratory, Pennsylvania State University. Stutz, B.,. Reboud, J. L. (1997). “Experiments on Unsteady Cavitation,” Experiments in Fluids 22: 191-198. Tulin, M. P. (1961). Supercavitating Flows, Handbook of Fluid Dynamics, V. L. Streeter, New York, McGraw Hill: 12.25 12.46. Werner, D. J. (1998). Technology Assessment of Hydrodynamic/Supercavitating Technologies, Alliant Techsystems Inc.

PAGE 72

62 BIOGRAPHICAL SKETCH Srihari Rasmi received his Bachelor of Technology degree from the Indian Institute of Technology, Madras, India, in 1999 in naval architecture. During December 1998 to April 1999, he received a scholarship to pursue his research interests related to stability of ships at Hochshule Bremen in Bremen, Germany. From August-December of 1999, he was an academic visitor at the Ship Stability Research Center at the University of Strathclyde in Scotland. Since January 2000, he has been pursuing his M.S degree in the Department of Aerospace Engineering, Mechanics, and Engineering Science at the University of Florida in the areas of experimental fluid dynamics and signal processing. 62


Permanent Link: http://ufdc.ufl.edu/UFE0000600/00001

Material Information

Title: Characterization of the University of Florida air-water shear-layer facility
Physical Description: Mixed Material
Creator: Rasmi, Srihari ( Author, Primary )
Publication Date: 2002
Copyright Date: 2002

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000600:00001

Permanent Link: http://ufdc.ufl.edu/UFE0000600/00001

Material Information

Title: Characterization of the University of Florida air-water shear-layer facility
Physical Description: Mixed Material
Creator: Rasmi, Srihari ( Author, Primary )
Publication Date: 2002
Copyright Date: 2002

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000600:00001


This item has the following downloads:


Full Text











CHARACTERIZATION OF THE UNIVERSITY OF FLORIDA AIR-WATER SHEAR-
LAYER FACILITY













By

SRIHARI RASMI


A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF
FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF MASTER OF SCIENCE


UNIVERSITY OF FLORIDA


2002















ACKNOWLEDGMENTS

I would like to thank my advisor, Dr. Louis Cattafesta, for his constant

encouragement and guidance. I also thank Dr. Mark Sheplak for his support and tutelage

efforts.

Mr. Kurt Banazynski of ELD Inc. was instrumental in the manufacturing process

of the University of Florida air-water shear layer facility. I also thank him for his

promptness in attending to matters related to the facility.

I greatly acknowledge the presence of Chris Bahr who helped me out in setting up

and carrying out various experiments. I also thank Ryan Holman for kindly letting me

borrow his codes to work upon.

Financial support for this project was provided by the Office of Naval Research

under Grant number N0014-00-1-0105.
















TABLE OF CONTENTS

Page
A C K N O W L E D G M E N T S ................................................................................................... ii

LIST O F TA BLE S ................................. ............................... .................. v

L IST O F FIG U R E S ............................................................................ ......... vi

A B S T R A C T ................................ ....................................................... ix

CHAPTERS

1 IN TR O D U C TION ................................... ... ......... ................. ... 1
S u p ercav itatio n ................................................................. 3
M otiv action for F facility ........... ............................................................... .. ............ 6
2 DESIGN AND CONSTRUCTION .............................................................. 11
Six-inch Flow Visualization W ater Tunnel .................. ............... ........ ............ 11
Flow Sections ............................................ 12
T e st S ectio n ................................13.............................
P u m p .................. ...... .. ................................................................. 1 5
V ariable Speed D rive A ssem bly ........................................ ......................... 15
Piping and Supporting Framework ................................. ........ .............. 15
Six-inch High-Speed, Open-Circuit Wind Tunnel..................... .. ........... ... 16
Flow Ducts and Flow Straighteners .............. ...... ......................... ............ 17
Test Section ............. ............. .. .... .................... ........... 17
Primary Diffuser and Liquid-Gas Separator ..................................................... 19
F an A ssem b ly ................... ....... ............. .. ............................ 2 0
M otor and M otor C ontroller................................................. ........... .............. 20
Supporting Frame ...... ........ ............. ........... ..... ............ .. ........ .. 20
3 CHARACTERIZATION OF FACILITY.............................. ................. 22
Velocity Profiles of the Air Tunnel..................................................... .................. 22
Experim mental Set-up .............. .... .... .... ......... .... .. .............. ...... 22
Calibration of Air Tunnel Velocity and Air Pressure Transducer............................. 24
V velocity Profi les.................................. .. ............. 25
Velocity Profiles of the Water Tunnel .......................... ......... .................. 27
Experim mental Set-up .... ...... ...... ............... ..... ............... ..... 27
Calibration of Water Tunnel Velocity and Water Pressure Transducer ................ 28
V elocity Profi les..................................... ...................... .. .... ..... .. .............. 29










Vibration Measurements of the Test Section .................................. ............. 31
E xperim mental Set-up .............................. .. ..................................... ......... .. .. 3 1
Theoretical M odel .. ................. ................... ............ .......... .. 33
R results and C onclusions....................................................... ......................... 36
4 INTERFACE EDUCTION USING LIGHT SHEET ............................................. 44
E x p erim en tal S et-u p ............................ .. ................ .................................................. 4 4
P ro c e d u re ............................................ ............................................... 4 5
In terface D etectio n ..................................................................... ....... .............. 4 7
O u tlier R ejectio n ......... ... ...................................................... ............ ..... 4 9
Transform ation and C alibration ........................................ .......................... 51
Interface Slope Calculation ............................ ........ ........ .............. 54
CONCLUSIONS AND FUTURE WORK ....................................................... 56

APPENDIX

TUNNEL OPERATION PROCEDURE ........................................ ...................... 57

REFERENCES ................................. ............... 60

BIOGRAPHICAL SKETCH ....................................................... .............. 62






























iv















LIST OF TABLES

Table Page

3-1 A ir tunnel velocity characteristics ........................................ .................. ...... 27

3-2 W ater tunnel velocity characteristics.................................. .................................... 31

3-3 Test matrix for vibration measurement................................................................... 33

4-1 Critical parameters of the laser and camera............... ..................................... 45

4-2 Test m atrix for interface education .................................................................. ... ... 46

4-3 The maximum error between the data and polynomial fit............... .................. 54















LIST OF FIGURES


Figure Page

1-1 Tip vortex cavitation in a marine propeller ............... .................................... 2

1-2 Schematic of a supercavitating vehicle............................................... 3

1-3 Schematic of interfacial conditions in the air-water shear layer facility. ................. 6

1-4 Schematic of interfacial conditions of a supercavitating vehicle.............................. 7

1-5 Air-water test facility for sensor characterization................................ .... ............... 8

1-6 Velocity profiles of a supercavitating vehicle (left) and the air water shear layer
fa cility (rig h t). .............................................. ............................ 9

2 -1 Inlet plenu m an d pu m p ............................................................................................. 13

2-2 Return plenum and test section. ............................................ .................... ...... 14

2-3 Test section of University of Florida shear layer facility. .................. ................ 14

2-4 First part blower housing, wide-angle diffuser, flow straighteners and contraction
section ............................................................................................. 16

2-5 Primary diffuser and liquid gas separator. .................................................. ........ 19

2-6 Rubber-in-shear mounts to minimize vibrations........................ .................. 21

3-1 Experimental set-up used to measure velocity profile in air tunnel ........................ 23

3-2 Schematic of the pitot-static probe mounting in the test section. ........................... 23

3-3 Motor frequency and velocity of air tunnel plotted as a function of DAQ input
v oltage. .............................................................................. 2 5

3-4 Block diagram for calibration and velocity profile measurement for the air tunnel.. 25










3-5 Velocity profiles (with error bars) of the air tunnel for velocities increasing from 6
m/s to 66 m/s in ten steps (corresponding to motor input voltage increasing from 1 V
to 10 V in step s 1 V ).... ......... .................. ............ .................... .. .................. 2 6

3-6 Motor frequency of water tunnel plotted as a function of DAQ input voltage.......... 29

3-7 Block diagram for calibration and velocity measurement for the water tunnel......... 29

3-8 Velocity profiles (with error bars) of water tunnel for velocity increasing from 0.15
m/s to 0.86 m/s in seven steps (corresponding to motor input voltage increasing from
1 V to 8 V in steps of 1 V). ......... ................................................................... 30

3-9 Accelerometers mounted at possible vibration sources, i.e. at the (a) blower and the
(b) pum p respectively .................................................................. ..... ... .... .. 32

3-10 Accelerometer mounted approximately at the midpoint of the test section to detect
vibration levels. Flow direction is along Y ............. ............................ ....... ....... 32

3-11 Power spectra of three components for all three accelerometers............................. 34

3-12 Two-input single output model for vibration analysis ......................................... 34

3-13 Top two plots show the ordinary coherent output spectrum due to the inputs. The
bottom two plots compare the noise spectrum with the output spectrum, the multiple
coherent output spectrum and the noise floor. .............................................. 38

3-14 The topmost plot shows r2, the coherence between the inputs. The next two plots
are 1, and 1z the ordinary coherence functions between each input and the output.
The bottommost plot is yx, the multiple coherence function.................................. 39

3-15 Comparison of output spectrum with BK microphone measurements. In the top
plot, the microphone is placed very close to the blower, with the pump off. In the
bottom plot, the microphone is placed very close to the pump, with the blower off.41

3-16 Comparison of the output spectrum with microphone measurements for the case
when the fan is off and the water tunnel speed is 0.5 m/s............... .................. 42

4-1 Experimental set-up with Nd:YAG LASER and PIVCAM ...................................... 45

4-2 Sample original image (left) and corresponding output from edge detection algorithm
(right). The flow direction is from left to right as indicated. ................................. 47

4-3 A sample edge image and the corresponding interface image as a result of pruning.48










4-4 Average interface and standard deviation of air-water interface. Number of statistics:
2 0 0 .................... ..... ... .................................. ................................ 4 9

4-5 Superposition of sequence of edge images before outlier rejection and a column (at
188th X pixel) of sum of sequence of interface images with an outlier point. ......... 49

4-6 Sum of sequence of edge images after outlier rejection and a column (at 188th X
pixel) of sum of sequence of interface images without any outlier points ............ 50

4-7 Comparison of interface images without and with outlier rejection. The average
images are shown before outlier rejection (left) and after outlier rejection (right)... 51

4-8 Comparison of standard deviation values before and after outlier rejection ............ 51

4-9 Projection transformation and reverse projection transformation of the calibration
im age ..... ........................... ...................... ....... .......... ..... 52

4-10 The average interface image and the corresponding reverse transformation to world
coordinates. .................... ........................ ............... 53

4-11 Interface image in world coordinates and corresponding polynomial fit ............... 54

4-12 Polynomial fit, of order 5, of average interface image (left) and corresponding slope
of interface (right). ..................... ............................................................. ......... ...... 55















Abstract of Thesis Presented to the Graduate School of the University of Florida in
Partial Fulfillment of the Requirements for the Degree of Master of Science


CHARACTERIZATION OF THE UNIVERSITY OF FLORIDA AIR-WATER SHEAR
LAYER FACILITY

By

Srihari Rasmi

December 2002

Chairman: Dr. Louis N. Cattafesta III
Major Department: Mechanical and Aerospace Engineering

Technological advancements in many fields have contributed to envisioning

supercavitating vehicles that can travel at far greater speeds than current underwater

vehicles. Under supercavitating conditions, the vehicle is enveloped in a single gaseous

bubble also known as a supercavity. Realizing such a vehicle admits many possibilities

for research in related fields. One of the chief issues is monitoring the cavity thickness to

control the vehicle. For this purpose, new algorithms and sensors have been developed

apart from performing comprehensive numerical simulations, etc.

For testing the sensors and algorithms, a novel facility has been designed and

constructed. The facility is an air-water shear-layer system that consists of an air and a

water test section with a free surface. The integral test section is 48 in. long by 6 in. wide

by 12 in. high. The air stream speed can reach a maximum of approximately 75 m/s,

corresponding to the approximate speed of a supercavitating vehicle. The water section

is designed with a maximum speed of approximately 0.8 m/s, which is high enough to










produce natural cavitation bubbles. The interfacial instabilities arising at the free surface

allow for sensor characterization and algorithm testing in an environment that simulates

supercavitation while maintaining low cost.

In this thesis, some of the more important features such as the velocity profiles of

air and water streams, vibration characteristics and interface conditions are studied.

Velocity profile measurements of both air and water tunnels are carried out using pitot-

static probes. The velocity profiles of the air tunnel are approximately uniform

throughout the test section outside of the boundary layer. For the water tunnel, large

error bars lead to inconclusive results regarding the mean flow quality in the test section.

Vibration measurements are made using accelerometers. This is to deduce the influence

of vibration from fan and pump (driving the air and water tunnels) on the flow quality

needs as a part of future work. An interface education method is developed to extract the

interface from series of flow visualization images for a particular interface condition.

Statistical quantities related to interface shape are obtained and discussed.















CHAPTER 1
INTRODUCTION

Realizing high speeds in underwater vehicles has been a challenge mainly because of

propulsion limitations and skin friction drag. The problem of skin friction drag is

compounded by the fact that it increases with the square of the velocity of the vehicle.

However, the last half-century has seen significant advancement in technological

concepts that makes achievement of greater speeds a definite possibility for underwater

vehicles [Tulin 1961]. Such a possibility has been envisioned as a result of a

phenomenon known as supercavitation. Under supercavitating conditions, an underwater

vehicle is entirely encapsulated in a large gaseous bubble or cavity. Realizing this

technology requires a thorough understanding of the physics of the problem. In addition,

control and maneuverability of the vehicle pose new challenges because of the different

dynamics involved. In this situation, the measurement of the cavity shape with

appropriate sensors is imperative. The characterization of these sensors demands an

innovative test bed that is cost effective and, at the same time, able to simulate the

supercavitation environment to a fair degree. This thesis aims at summarizing the

development, design and characterization of such a test bed for sensor testing specific to

supercavitating vehicles. The need for a unique test facility is built up from the definition

of cavitation as given below.

Cavitation is defined as the gas-liquid region formed due to pressure reduction as a

result of the dynamic action of the fluid on the boundaries of a liquid system [Stutz and








2

Reboud 1997]. This pressure reduction can be roughly quantified in the following way.

Bernoulli's equation along a streamline with negligible height change is


+ 1 v2 = constant (1.1)
p 2

where p is the pressure, p is the density and v is the speed of the liquid. When v

increases sufficiently so that p drops below the vapor pressure p, of the liquid, the

formation of bubbles, also known as cavities that contain the vapor of the liquid occurs.

This is cavitation. Based on this, the cavitationn number" a is defined as


S2(p (1.2)
pv2

The cavity can be filled with the vapor from the ambient liquid, as in the case of the

water vapor bubbles at the tip of a propeller blade in tip-vortex cavitation shown in

Figure 1-1 [Billet 2000]. However, a cavity can also be filled with any other gas. For

example, the formation of cavities of air in the rear of any body during water entry is very

common. Albeit the latter case does not adhere to definition of cavitation, the physics is

similar. Hence, such cavities are referred to as ventilated cavities [Tulin 1961].



:..:....: ...


Figure 1-1 Tip vortex cavitation in a marine propeller (from Billet 2000).








3

Cavitation is usually very localized and for the most part is known to have

detrimental effects. For example, in bubble cavitation of marine propellers, the rapid

formation and collapse of bubbles can be very harmful to the blade of the propeller [Stutz

and Reboud 1997]. As a result the blade suffers from fatigue and loss of material. Hence

a great incentive lies in predicting cavitation and studying the effect of cavitation with

respect to marine propellers. In general, cavitation occurs in various environments like

low head pumps, heart valves, venturi meters etc. To model these, analytical methods

that employ potential theory have been used with a degree of success [Kunz et al. 2000].

However, cavitation often includes mass transfer, unsteadiness and viscous effects [Stutz

and Reboud 1997; Kunz et al. 2000].

Supercavitation

Supercavitation is a specific cavitating condition wherein a vehicle moving in a

liquid is completely enveloped by a gaseous cavity. In a supercavitating condition, only a

single gaseous bubble exists that is usually much larger than the body itself and stays

attached to the body at all times [Tulin 1961]. A schematic of a supercavitating vehicle is

shown in Figure 1-2 [Kam Ng 1999].


Cavitator Supercavity

............:.... ........... .. X .4 ......
.:..:.::.:.....::... ::. ...... ................ .. ;::..i .. .iii.iii::iiiiii::i :: ii: iiiii::i :: .: : .::i .



Figure 1-2 Schematic of a supercavitating vehicle [Kam Ng 1999].

The incentive to employ supercavitation in underwater vehicles is that the skin

friction drag is reduced to a great extent because the viscosity and density of the gas is








4

considerably less than that of the liquid [Werner 1998]. This leads to greater thrust and

thus greater speeds. High-speed supercavitating vehicles can travel at hundreds of knots

[Tulin 1961; Kulkarni and Pratap 1999].

Supercavitation can take two forms. First, natural cavitation occurs when the

vehicle is traveling fast enough to vaporize the ambient liquid and form a single bubble

cavity. Second, artificial cavitation or ventilation occurs when a ventilator is present at

the nose of the body to inject gases and form a cavity (Figure 1-2). In the second method,

which is more common, the exhaust gases are fed through the ventilator. Irrespective of

natural or artificial inception of the cavity, the term supercavitation is applied for both

cases [Tulin 1961].

The optimal design of a supercavitating vehicle results in significant drag

reduction, since only small regions at the nose and afterbody control surfaces contact the

water. Design considerations in such vehicles mainly include the shape of the cavitator,

vehicle size and shape, shape of fins, and dynamics of the vehicle. The cavitator dictates

the shape and size of the cavity. Cavitator shapes in the form of disks, cones, etc. have

been experimented with [Stinebring et al. 2000, 2001]. The small size of the cavitator

with respect to the body ensures minimal pressure drag for the speed range of interest.

With the size of the cavity fixed by the shape and orientation of the cavitator and the

given speed range, the size of the vehicle itself is restricted as a result of this. The

vehicle houses the propulsion system and control and guidance system among other parts.

The fins are the primary control surfaces that mostly cut well into the cavity wall (Figure

1-2). The dynamics of the problem are increasingly complicated when the vehicle turns








5

wherein tail slap at the cavity wall is a definite possibility [Kulkami and Pratap 1999].

Vehicle rotation has also been predicted [Kulkami and Pratap 1999].

The aforementioned design considerations and conditions have led to piecewise

research in specific areas like cavitator designs [Stinebring et al. 2001], numerical

prediction [Kunz et al. 2000], cavity thickness measurement using novel methods [Li et

al. 2002a, 2002b, 2002c] etc. However, supercavitation itself has been studied for a long

time. The use of supercavitation in high-speed marine propellers, hydrofoil boats, and

low-head pumps/turbines sparked a renewed interest in the subject [Tulin 1961]. Added

to this, supercavitation with respect to bodies of revolution have been under scrutiny in

recent times especially with regard to underwater artillery [Tulin 1961]. As a result,

associated topics such as water entry, dynamics, and control have been the subjects of

theoretical and experimental studies [Tulin 1961].

While supercavitation applied to underwater vehicles is of a great tactical merit,

one of the principal tradeoffs is the stability and maneuverability of the vehicle. While

the stability is dictated by the design of the vehicle and also by the dynamics of the

cavity, the maneuverability depends on the prediction and control of the cavity shape and

vehicle dynamics in the presence of fins. For the prediction and control of the cavity,

there is a need to (a) measure the local cavity thickness, (b) understand the interfacial

instabilities at the cavity wall and (c) employ a suitable control algorithm. The

measurement of cavity thickness has necessitated the need for innovative technology.

Related research areas including algorithm development, sensor manufacturing and

experimentation are in progress [Chandrasekharan et al. 2001a, 2001b, 2002; Li et al.

2002a, 2002b, 2002c].








6

Motivation for Facility

For the new algorithms and sensors that are currently being developed, a cost

efficient testbed that mimics supercavitating conditions is required for experimentation.

This provides the primary motivation for the University of Florida air-water shear-layer

facility. This is because setting up a conventional supercavitating environment in lab

conditions is a challenging and costly problem. Such a set-up would require a test section

in which water flows at high speeds, even to maintain a supercavity under ventilating

conditions. To circumvent the complexity and cost factor involved in such a project, an

affordable solution is sought in the form of an air-water shear layer interface (see Figure

1-3). A supercavitating environment is mimicked by keeping the air speed at

approximately 150 knots and maintaining the speed of water to be high enough to

produce cavitation bubbles (approximately 1 m/s) [Sheplak et al. 1999]. This system is

certainly a low cost option as conventional low speed air tunnels and water tunnels can be

manufactured at a much lower cost than a custom water tunnel that has to propel water at

high speeds.



I ) Air: -76 mis
Mean water
splitterr plate level







Figure 1-3 Schematic of interfacial conditions in the air-water shear layer facility.

To assess the ability of an air-water shear layer facility to simulate the actual

supercavitating conditions, a comparison of the air-water interface conditions in both







7

cases is now addressed. For a supercavitating vehicle, a schematic of the cavity interface

is shown in Figure 1-4.







T 1 ____-. \
-, P -- -'-\ .





U _._____--__-_-_



Figure 1-4 Schematic of interfacial conditions of a supercavitating vehicle.

The various stages of the interface along the length of the underwater vehicle are

illustrated. Interfacial instabilities that are initiated at the nose of the vehicle propagate

towards the aft of the vehicle, growing in amplitude. Hence, while the cavity near the

nose of the vehicle is minimally unsteady, the closing region of the cavity suffers

considerable interfacial unsteadiness. At the nose, the cavity surface is more or less

glasslike. Typically very few gas bubbles exist in this region. The next stage shown

indicates the amplification of instabilities initiated at the nose of the vehicle. Near the aft

of the vehicle, the air-water interface may break down, leading to very high drag forces.

This region may contain high concentrations of water vapor. Also, there is significant

mixing of the two fluids. For comparison, a schematic of the air-water interface of the

air-water shear layer facility is shown in Figure 1-3. A splitter plate separates the air and

water streams. Much like in a supercavitating vehicle, the instabilities begin at the

splitter plate and propagate along the flow direction. The region near the splitter plate








8

suffers very small disturbances whereas towards far downstream of the test section,

amplification results in significant disturbances.

The shear-layer facility is established as a competent substitute to actual

supercavitating conditions. To test distance estimation algorithm and sensors, a sensor

package housed in a flat plate is inserted in the air stream of the air-water shear layer

facility as shown in Figure 1-5. A flat plate that houses the sensor package with an

embedded sensor package is inserted into the high-speed air stream. Case studies, by

varying the position x, the orientation 0, and height h of the sensor above the interface,

are possible for various conditions of the air/water interface.

x

air I- I t
1, 1 171i .t-
U air


h


I. lE I.:- .. : i

,t ii:., .- flat plate who e

_______________________________________ angle 1
I,-: .il l. I .Il : -.' II,-
air-water interface
is variable


Figure 1-5 Air-water test facility for sensor characterization.

The main drawback of the air-water shear-layer facility is that it does not exactly

match the operating conditions of a high-speed supercavitating vehicle. The difference

arises from the non-moving flat plate, which leads to different velocity profiles between

the two cases. This is shown in Figure 1-6. With the supercavitating vehicle traveling at

high speed, the gas adjacent to it is at the same speed in accordance with the no-slip








9

boundary condition. Hence, the velocity monotonically asymptotes to zero while

maintaining continuity at the interface. Whereas in the air-water shear-layer facility,

since the plate is stationary and the free stream speed is nonzero, a boundary layer

develops on the flat plate. The state of the boundary layer that develops and its state can

be controlled by proper selection of the plate length and/or a boundary-layer trip [Sheplak

et al. 1999]. From the plate, the velocity increases from zero, in accordance with the no

slip boundary condition, to the free stream velocity and then asymptotically drops to the

water stream speed. Notwithstanding this difference, the air water shear layer facility

does simulate many of the important features of the cavity interface of a supercavitating

vehicle.


Air Air


Adju stible
height
Free stream


vehicle





Figure 1-6 Velocity profiles of a supercavitating vehicle (left) and the air water shear
layer facility (right).

The facility described in Chapter 2 is fabricated by Engineering Laboratory

Design, Inc. The salient features of this facility are 1) a plexiglas test section with

dimensions 6 in. wide by 12 in. high by 48 in. long (a 6 in. square section for each








10

stream), 2) a maximum air speed of approximately 76 m/s, and 3) a maximum water

speed at just under 1 m/s which is high enough to produce natural cavitation bubbles.

The next chapter will focus on a detailed design of the air water shear layer

facility along with description of the chief components used.















CHAPTER 2
DESIGN AND CONSTRUCTION

The University of Florida Air-Water Shear Layer Facility is a unique testbed

developed for sensor characterization and algorithm testing related to supercavitating

vehicles. This chapter is relies heavily on information presented in Banazynski [2000].

The facility has overall dimensions of 27 ft. long by 3.92 ft. high by 6.2 ft. wide and

a net weight of 1900 lbs. The test section (Figure 2-3) measures 48 in. in length by 6 in.

in width by 12 in. high. The centerline of the test section is approximately 62 in. from

the ground.

The facility consists of three principal parts that can be readily joined/disjoined. The

first part houses the inlet, blower, contraction section, etc. The second part comprises the

closed-loop water tunnel inclusive of the test section and water pump. The third part is

the primary diffuser of the wind tunnel with a liquid gas separator unit. These three parts

are carefully aligned and joined by screws.

The shear-layer facility can be classified according to its two main components: the

water tunnel and the air tunnel. The design and parts of the water tunnel are described in

detail below followed by the same for the air tunnel.

Six-inch Flow Visualization Water Tunnel

The water tunnel is an un-pressurized system that is designed to operate in a closed-

circuit configuration a vertical flow loop. It consists of the pump, flow sections,








12

supporting framework and the test section. The test section may be operated as a

conventional water tunnel with a plexiglas cover in place.

This unit is 10.375 ft. long by 5.58 ft. high by 2.5 ft. wide. The test section

centerline height is 4.52 ft. from the ground. The gross weight of the system, filled with

water, is approximately 660 lb. The tunnel can hold approximately 48 gallons of water.

The overall construction and different components used in the water tunnel are

described below.

Flow Sections

All elements of the system in contact with water are fabricated from non-corrosive

materials. The flow sections are fabricated from laminated fiberglass reinforced plastic.

This material is used because it is strong and lightweight. It easily withstands both

applied static and dynamic loads. Interior surfaces have a glass smooth, white, vinyl-

ester, gel-coat. Exterior surfaces are spray finished with a medium blue polyester and

gel-coat enamel. To ensure water tightness, flanged joints are sealed using a high quality,

polyurethane marine adhesive/sealant. All joints are secured using stainless steel

fasteners. Any discontinuities in fit and alignment could lead to disturbances in flow.

Hence, careful attention is paid in joining mating sections. The covers for the fiberglass

ducts are fabricated from 0.5 in. thick plexiglas.

The inlet plenum (Figure 2-1), also called the distributed plenum, and the return

plenum are adjacent to the test section on either side. A perforated cylinder distributes

flow from a centrifugal pump into the diffusing part of the inlet plenum. Perforated

plates of stainless steel act as head loss baffles in the diffuser. For a settling chamber, a

precision, tubular cell, plastic, honeycomb section is inserted just before the contraction.








13

Adjacent to this, two stainless steel screens (60% porosity) are mounted. Both the

honeycomb section and stainless steel screens act as flow straighteners. A removable

cover provides access to this area.

In the return plenum (Figure 2-2), twin-turning vane cascades are placed to direct the

flow leaving the test section. A removable plexiglas end wall enables observation from

downstream.

Test Section

The test section sides and flanges are made of 0.5 in. thick, clear acrylic plexiglas.

The test section floor is made of 0.75 in. thick, clear, acrylic plexiglas. The test section

interior is 48 in. long by 6 in. high by 6 in. wide (Figure 2-3). A portion of this excess

height is used as the sidewall boundary for the wind tunnel test section.


Figure 2-1 Inlet plenum and pump.





























Figure 2-2 Return plenum and test section.


1w


*---u a--4L~


Figure 2-3 Test section of University of Florida shear layer facility.

To operate the water tunnel in a conventional water-tunnel only mode, a cover

fabricated from V2 in. thick, clear acrylic plexiglas is provided. The cover has lap flanges








15

that correspond to steps machined into the permanent mid-height ends in the test section.

The cover is secured to the mid-height ends with nylon screws.

Pump

A stainless steel centrifugal pump (G&L SSH Model 6SH2F2GO, with a 5.625 in.

diameter impeller) is provided to circulate the water in the tunnel at a desired speed. The

pump delivers 280 GPM using 1.5 HP. The pump is directly driven by a 1.5 HP, open

drip-proof, 1800 RPM, 208/230 VAC induction motor (Baldor International Model No.

JMM3154T).

Variable Speed Drive Assembly

The pump shaft RPM (frequency) is controlled by a transistor inverter type, variable

frequency inverter (Toshiba Model No. VFS7S-2015UP). The inverter is arranged for

208/230 VAC, single-phase input electric service. A remote control station, located

adjacent to the test section, provides a user-friendly interface to regulate the test section

velocity.

A NEMA rate fusible disconnect is wired inline to protect both the motor and

controller. The disconnect, inverter and related components are mounted in a NEMA

Type 4 enclosure.

Piping and Supporting Framework

The majority of the piping is commercial PVC pipe and fittings. Flexible rubber

couplings are used to join the piping to the flow sections to provide water tightness and to

act as vibration dampers. A drain valve is provided at the lowest point in the system.

The supporting framework consists primarily of structural steel tubing. The

framework consists of two parts: the main frame and the pump frame. The main frame








16

holds the flow section and test section. The main frame is fitted with swivel casters for

transportation. The pump frame supports the pump to isolate the test section from pump

vibrations. Each frame leg is fitted with an adjustable leveling pad. The pump frame is

carried on an extension of the main frame. The frames are etched, prime coated and

spray finished with acrylic enamel.

Six-inch High-Speed, Open-Circuit Wind Tunnel

The wind tunnel system is an open-circuit blower-type system. Air is sucked normal

to the flow direction into the inlet by the fan as shown in Figure 2-4. The system consists

of flow sections and straighteners, test section, primary diffuser and a liquid-gas separator

unit. The unit is 27.2 ft. long by 6.29 ft. high by 3.45 ft. wide. The total weight of the

unit is approximately 1250 lbs.

LOW STRAIGHTEE R


Sir intake





CONTRACTION 9:1 '
IDE ANGLE
DIFFUSER








Figure 2-4 First part blower housing, wide-angle diffuser, flow straighteners and
contraction section.

The overall construction and different components used in the wind tunnel are

described below.










Flow Ducts and Flow Straighteners

The flow ducts are fabricated from fiber reinforced plastic molded over precision

tooling. Interior surfaces have glass smooth, white, vinyl-ester, gel-coat, while exterior

surfaces are spray finished with a medium blue, polyester, gel-coat enamel.

The system air from the blower is discharged through a wide-angle diffuser that has

perpendicular lap flanges. Then face flanges are fastened with bolts. The wide-angle

diffuser section expands with a total included angle of 12.3 in the horizontal plane and

1.4 in the vertical plane. The wide-angle diffuser helps in lowering the speed before the

air enters through the flow straighteners. In the settling length, a precision, hexagonal

cell, aluminum honeycomb section is placed. This is followed by a mesh of high porosity

(60%) stainless steel screens. The screens are mounted and tensioned using a proprietary

design. Extruded aluminum frames cover the entire settling length. The settling length is

followed by the contraction section that is a single-piece mold. The cross section of the

contraction section is symmetric throughout, has a 9:1 ratio, and has analytically

developed contours. The contraction section leads to the test section through a transition

plexiglas section. Flow from the test section continues into the diffuser section that has

miter flanges. The primary diffuser section expands with a total included angle of 7.

High porosity perforated plates are fastened 14.81 in. from the diffuser entrance and at

the diffuser exit. Any discontinuities in fit and alignment could lead to disturbances in

flow. Hence, careful attention is paid in joining mating sections.

Test Section

The test section is integral to the water tunnel test section. The water tunnel test

section is 48 in. long by 6 in. high by 6 in. wide in the interior. The total test section is a








18

single plexiglas mold that is 48 in. long by 12 in. high by 6 in. wide. The removable test

section is joined to the plexiglas transition section and the diffuser ducts with flanges.

The wind tunnel cover consists of a hinged ceiling plate that can be raised or lowered

with a screw. As a result, the ceiling can be diverged from 0 to 1.80. To minimize air

leak at the sides of the ceiling plate, a self-adhesive felt gasket is used as sealant. The

ceiling for the duct downstream from the test section is connected to the diverging ceiling

and transitions back to the 6.00 in. test section height. Along the span wise centerline of

the ceiling plate, a 0.63 in. wide by 43.18 in. long slot is machined on the diverging part.

The slot allows access of instrumentation into the test section. To minimize air leaks

through the unused area of the slot, a high-density nylon brush with extruded aluminum

holder is used. This brush is secured to the ceiling plate with screws. A series of 22

static pressure taps (with nylon barbed fittings) are located at 1.50 in. distance from the

spanwise centerline towards the sidewall. The centers of the static pressure ports are

placed at an interval of 2.13 in. along the flow direction.

To run the test section as a conventional air wind tunnel, a removable floor,

fabricated from 0.5 in. thick, clear acrylic plexiglas is provided. The floor has lap flanges

that correspond to steps machined into the permanent mid-height ends in the test section.

The floor is secured with screws.

An additional, shorter wind tunnel test section is provided to allow the wind tunnel to

be used separately from the water tunnel. The test section can be mounted directly to the

contraction exit. The working section is 18 in. long by 6 in. high by 6 in. wide in the

interior. Plexiglas flanges join the test section to the contraction section and the diffuser








19

ducts. A 15 in. portion of the ceiling is removable and secured with quick release

fasteners.

Primary Diffuser and Liquid-Gas Separator

The primary diffuser (Figure 2-5) serves to regain static pressure. It is a single-fiber

reinforced plastic mold that is joined with screws to the return plenum. The primary

diffuser section expands with a total included angle of 7. Flow continues through a

liquid/gas separator unit and is discharged to the atmosphere. A liquid-gas separator is

necessary because operation of the facility at air speeds greater than 40 m/s results in a

significant loss of water from the shear layer interface due to entrainment by the airflow.

The liquid/gas separator is a commercial model Harrington Model No. HPE-30. The

liquid/gas separator is installed at the farthest point downstream of the wind tunnel

diffuser. The water reclaimed by the mist eliminator can be discharged via a drain or

routed back to the water tunnel using a commercial pump.

-E- M _=


Figure 2-5 Primary diffuser and liquid gas separator.










Fan Assembly

A single width, single inlet (SWSI) centrifugal fan (TCF/Aerovent Model No. 16

BIA-SW-ARR: 10 CL2 CW THD) is used. The fan is arranged in a top horizontal

discharge configuration in which the inlet is perpendicular to the direction of airflow.

When viewed along the direction of flow, the air intake is from the left.

Motor and Motor Controller

The fan is belt driven by a 7.5 HP, ODP, 3600 RPM, 208-230 VAC, 3 4, 60 Hz

motor (Toshiba Model No. BY752LF2UMH04).

A transistor inverter type, variable speed motor control (Toshiba Model No. VFS9-

2055PL-WN) controls the fan shaft RPM. The controller is arranged for 208-230 VAC, 3

), 60 Hz input electric service. A remote control station, located adjacent to the test

section, allows easy regulation of the test section speed.

A NEMA rate fusible disconnect ensures the motor and controller are safe. The

disconnect is mounted in a NEMA Type 4 enclosure.

Supporting Frame

The wind tunnel is supported by structural steel tubing frames that are welded

together. The frames are etched, prime coated and spray finished with acrylic enamel.

Leveling pads and swivel casters are fitted for height adjustment and transportation. To

minimize conduction of vibration from room and fan/pump, rubber-in-shear mounts are

provided (Figure 2-6).








21











Figure 2-6 Rubber-in-shear mounts to minimize vibrations.

The next chapter will describe measurement velocity profiles for both the air and

water tunnels. Also, measurement of vibration characteristics of the test section as a

result of the fan and pump is described.















CHAPTER 3
CHARACTERIZATION OF FACILITY

To characterize the wind and water tunnel test sections, the mean flow quality of both

the sections and the influence of vibration of the fan and pump on the test section

vibration are considered here. The mean velocity profiles of both the test sections are

measured separately. For the velocity measurements, a pitot-static probe is used in

conjunction with a differential pressure transducer. Vibration measurements are carried

out by placing accelerometers at tentative vibration sources, namely the fan and the

pump, and the mid region of the test section. The velocity profile measurement for the

air tunnel is described below.

Velocity Profiles of the Air Tunnel

Experimental Set-up

The experimental set-up for measuring the velocity profile of the air stream is shown

in Figure 3-1. The pitot-static probe is inserted from the slot in the ceiling of the test

section at a distance of 19 in. from the start of the test section. The mounting of the pitot-

static probe and the coordinate system defined are shown in Figure 3-2. To avoid the air

water interface, the tunnels are isolated using a plexiglas cover. Along with the pitot-

static probe, two pressure transducers, Validyne P855A and a Heise HSQ-1 (Range: 0 -

15 in. H20), are used to measure the dynamic head of the flow. A T-junction is used to

split the pressure lines from the pitot-static probes to connect to the Heise and the

Validyne transducers. The voltage output from the Validyne P855A, in the range from -

5 to +5 V DC, is read using a HP 34970A integrating voltmeter.

22


















































Figure 3-1 Experimental set-up used to measure velocity profile in air tunnel.


Pitot
static
probe












18.0 in.


Figure 3-2 Schematic of the pitot-static probe mounting in the test section.








24

Additional apparatus included a micron resolution 3D traverse system for precisely

positioning the pitot-static probe and an NI PCI6024E series DAQ card with

programmable voltage output for controlling the speed of the air tunnel. The output

voltage from the DAQ card transforms to a frequency of operation for the motor that

operates the fan for the air stream. A LabVIEW program is used to integrate 1) data

acquisition from the HP 34970A DAQ unit over a GPIB interface, 2) data acquisition

from the Heise pressure gauge with a serial port connection and 3) control the NI

PCI6024E output voltage, to set the air speed. The operating instructions for this

LabVIEW program as well as the tunnel operating instructions are explained in

Appendix A.

Calibration of Air Tunnel Velocity and Air Pressure Transducer

The pitot-static probe is fixed at the mid-height of the tunnel test section. The block

diagram for calibration is given shown in Figure 3-3. In an automated process, the input

voltage, va, to the motor is increased from 0 10 V DC in steps of 0.2 V. For each va, 25

readings are taken from the Heise gauge and the HP 34970A DAQ unit over a period of

approximately 10 seconds. These 25 readings are averaged to obtain a single reading

from the Heise Gauge and HP 34970A DAQ unit for that particular v,. The result

recorded achieves calibration of the wind tunnel speed with the DAQ voltage input and is

used for all subsequent measurements. This is shown in Figure 3-4. A look-up table for

Va versus the motor frequency is recorded to facilitate manual setting of desired speed

based on frequency. The dependence is linear except at very low input voltage. From

this, the threshold value of the input is found to be approximately 0.6 V.




















Figure 3-3 Block diagram for calibration and velocity profile measurement for the air
tunnel.




70 70

60 ------ -- ------ A60
50 -------- ------ ---- ------- -50-
50

E
cc540 -------- ------- -------- -- --- --" ----50-

A >


S20 20
620 --------------------- ------- -------120

- ---- - ----- - ----0
10 2, 10


0 2 4 6 8 10
DAQ voltage (Volts)


Figure 3-4 Motor frequency and velocity of air tunnel plotted as a function of DAQ input
voltage.

Data from the HP 34970A DAQ unit (output voltage from the Validyne P855A) is

used to calibrate the P855A with respect to the Heise gauge.

Velocity Profiles

The block diagram for the velocity profile measurements is shown in Figure 3-3. For

the velocity profile measurement, the pitot-static probe is traversed from the bottom of

the tunnel to the top of the tunnel in steps of 2 mm while keeping va constant. Then, va is

steadily increased from 0 10 V DC in steps of 0.2 V. An average of 25 readings from








26

the Heise gauge are recorded for each v, and each traverse position. The results are

plotted as velocity profiles as shown in Figure 3-5. The velocity profiles do not reach

zero at the floor because of the finite diameter (} in.) of the pitot-static probe.


70 1


60 4 ---- --- --- --- --- --- --- -
60 .




0 .......





0 _
01 2 3 4 5 6


Tunnel profile (inches)
0 ------------------------^-----f--~ Ct~1



---,. F = =_ == = ._. = -.. .= -. ^ '- -

FPoor 1 2 3 4 5 6

Tunnel profile (inches)


Figure 3-5 Velocity profiles (with uncertainty estimates) of the air tunnel for velocities
increasing from 6 m/s to 66 m/s in ten steps (corresponding to motor input voltage
increasing from 1 V to 10 V in steps 1 V).

Table 3-1 summarizes the results of the velocity profile measurements for the air

tunnel. The mean, maximum percent absolute deviation and standard deviations are

tabulated outside the local boundary layer. At a particular operating condition, the

maximum absolute deviation is calculated as the largest absolute magnitude of the

difference between the velocity measurements and the mean value divided by the mean

value. This deviation is found to reduce as the mean speed increased while the standard

deviation values increased progressively.










Table 3-1 Air tunnel velocity characteristics

Mean % Maximum % Standard deviation
velocity (m/s) absolute deviation (with respect to mean)
6.1 2.17 0.98

12.4 0.76 0.32

19.0 0.70 0.31

25.7 1.19 0.47

32.4 0.82 0.34

38.9 0.75 0.39

45.4 0.80 0.42

51.7 0.75 0.35

58.0 0.91 0.33

64.2 0.70 0.28


Velocity Profiles of the Water Tunnel

Experimental Set-up

The experimental set-up used here is similar to that of the air measurements except

that a free air-water surface is maintained in this case. The pitot-static probe is inserted

from the slot in the ceiling of the test section at a distance of 10 in. from the start of the

test section. The tubes from the probe are connected to the transducer after flooding all

parts with water. Visual inspection insured no bubbles were present. The pitot-static

probe is connected to Validyne P55D pressure transducer that has a full-scale range 2.5

in. of H20. The output of this transducer is connected to the HP 34970A DAQ unit. The

3D traverse with servomotors is used to position the probe. An NI PCI6024E series data

acquisition card with programmable voltage output is used for controlling the frequency

of the water pump, thereby controlling the speed of the water tunnel. An integrated








28

LabVIEW program is used to 1) control the water tunnel speed, 2) collect data from the

HP Unit using GPIB interface and 3) control the traverse. The operating instructions for

this LabVIEW program is given in Appendix A.

Calibration of Water Tunnel Velocity and Water Pressure Transducer

The pitot-static probe remains fixed at the center of the air tunnel test section. Access

to the air water interface is blocked by placing a plexiglas cover over the water tunnel.

The block diagram for calibration is shown in Figure 3-6. Input voltage Va to the motor is

increased in small intervals from 0 until a point where the dynamic head of the air stream

is a little less than 2.5 in. of H20. For each Va, 25 readings are taken from Heise Gauge

and also from HP 34970A DAQ unit over a period of approximately 10 seconds. Each of

these 25 readings is averaged to get a single reading from the Heise Gauge and HP

34970A DAQ unit for that particular Va. The result recorded achieves the calibration of

the P55D with respect to the Heise gauge.

To calibrate the water tunnel velocity with respect to the motor frequency, the pitot-

static probe is positioned at the center of the water tunnel. The input voltage, vw, to the

pump motor is increased from 0-10 V DC in steps of 0.2 V. For each vw, 25 readings are

taken from the P55D using the HP 34970A DAQ unit over a period of approximately 10

seconds. Each of these 25 readings is averaged to get a single reading for a particular v,.

The result recorded achieves calibration of the water tunnel speed with the DAQ voltage

input and is used for all measurements in future. This is shown in Figure 3-7. A look-up

table for v, versus the motor frequency is recorded simultaneously to facilitate manual

setting of desired speed based on frequency. The dependence is linear except at very low









29


input voltage. From this, the threshold value of the input is found to be approximately 0.6


V.


Calibration


Figure 3-6 Block diagram for calibration and velocity measurement for the water tunnel.


60


50
IN

-r
,40
C-
(,

220
" 20
0

10


09


2 4 6
DAQ voltage (Volts)


0.6_



0.4 '



0.2



0
8


Figure 3-7 Motor frequency of water tunnel plotted as a function of DAQ input voltage.


Velocity Profiles

For the velocity profile measurement of the water tunnel, the pitot-static probe is


traversed from the bottom of the water tunnel until just below the air water interface in


steps of 2 mm. The voltage vw is steadily increased from 1-8 V DC in steps of 1 V. An


-------- -------- --




T-- --------
- --------.---.
.
7

---


r








30

average of 25 readings from the water pressure transducer using the HP 34970A DAQ

unit are recorded for each Vw and each traverse positions near the floor. The resulting

velocity profiles are shown in the plot in Figure 3-8. Large uncertainties are observed,

particularly at high speeds and probe positions, because of vibration experienced by the

probe due to the air-water interface.

For both air and water streams, more accurate measurements like particle image

velocimetry and Laser Doppler velocimetry are required to determine the basic flow

pattern. The resulting velocity profiles can be used to develop analytical models of the

air-water interface.






0.8

E 0.6 -


S0.4 TmTTTi^^--

0.2

0
0 1 2 3 4 5 6
Floor Tunnel profile (inches) Interface


Figure 3-8 Velocity profiles (with uncertainty estimates) of water tunnel for velocity
increasing from 0.15 m/s to 0.86 m/s in seven steps (corresponding to motor input voltage
increasing from 1 V to 8 V in steps of 1 V).

Table 3-2 summarizes the results of the velocity profile measurements for the water

tunnel. The mean, maximum absolute deviation and standard deviations are tabulated as

with the air measurements. The maximum absolute deviation is found to fluctuate








31

without any pattern the standard deviation values steadily increased as the mean velocity

increased.


Table 3-2 Water tunnel velocity characteristics

Mean velocity % Maximum % Standard deviation
(m/s) absolute deviation (with respect to mean
0.16 1.02 0.56

0.26 1.43 0.54

0.38 4.30 1.18

0.50 2.19 0.86

0.61 1.36 0.51

0.74 2.30 1.22

0.82 2.21 0.68

0.85 0.90 0.49



Vibration Measurements of the Test Section

The vibrations induced by the fan and pump, especially at high operating speeds, may

vibrate the test section to a reasonable degree. Hence, the vibration influence of the fan

and pump on the test section may become critical to the flow quality of both the air and

water stream in the test sections. Therefore, vibration measurements at select operating

conditions are performed.

Experimental Set-up

Three tri-axial accelerometers (Piezotronic Model 356A16) are used to measure the

vibration levels. Two accelerometers are mounted close to the source of the vibration,

i.e., at the fan and the pump as shown in Figure 3-9. A third accelerometer is placed

approximately at the midpoint of the test section as shown in Figure 3-10. The

accelerometers are mounted so as to measure accelerations in a coordinate system that is







32

irrespective of any accelerometer placement is defined as shown in Figure 3-9 and Figure

3-10. All accelerometers are attached using wax.













(a) blower (b) pump
Figure 3-9 Accelerometers mounted at possible vibration sources, i.e. at the (a) blower
and the (b) pump respectively.















Figure 3-10 Accelerometer mounted approximately at the midpoint of the test section to
detect vibration levels. Flow direction is along X.

A signal conditioner is used prior to acquiring the data from the accelerometers. A

HP E1433 16-channel 52 kHz digitizer is used for acquiring the data. Each channel has a

built-in programmable anti-aliasing filter whose cutoff span, fc, is determined by the

sampling frequency, f, as

f =f/2.56 (3.1)








33

The data is recorded for the operating conditions listed in Table 3-3. The frequency

values in Table 3-1 correspond to the inverter frequency that is used to set the speed of

the fan and pump impeller.

Table 3-3 Test matrix for vibration measurement

Air (Hz|ms ')
00100.0 30132.4 60164.5
Water (Hz ms___
0010.0 *

3010.5 *

6010.9 *


For all cases, the acceleration vector (i.e., all three orthogonal components) is

measured. The signal is sampled at 2048 samples per second. For a sampling rate of

2048, a usable span of 800 Hz results. A total record length of 393216 points is

collected. The record length is broken into 192 blocks of 2048 samples, resulting in a

frequency bin of 1 Hz, to perform analysis. A hanning window is applied to the data and

an overlap ratio of 75% is used. The estimate of random error with 192 averages is

0.072.

Theoretical Model

The system is mathematically modeled as a two-input / single-output system as

shown in Figure 3-11 [Bendat and Piersol 2000]. In Figure 3-11, xi(t) corresponds to the

accelerometer data from the fan, x2(t) corresponds to the accelerometer data from the

pump, n(t) is uncorrelated noise (with the inputs xi(t) and x2(t)) and y(t) is the vibration

measured at the test section. Since the tri-axial accelerometer gives accelerations in three

directions, x1, x2, and y can be only one of the three components. In order to choose the










34


component, the power spectra of the three components for all three accelerometers are


plotted, as shown in Figure 3-12 below.


n (t)

xl(t) Hf)


H, 0


Figure 3-11 Two-input single output model for vibration analysis.


FAN


:.A I'', .'


100 200 300 400

PUMP


I' ''


I ;, ':


100 200 300 400

TEST SECTION






I I I I




100 200 300 400
Frequency (Hz)


500 600 700 800


A .


500 600 700 800



z axis
x axis
5y axis








500 600 700 800


Figure 3-12 Power spectra of three components for all three accelerometers.


100




10-




1010
0


io2
10-2


10


106


10-8

1010
0


102


10-4


106


10

10-10
0








35

Based on Figure 3-12, the x component of accelerometer mounted near the fan is

chosen as x1, the y component of accelerometer mounted near the pump is chosen as x2,

and the x component of accelerometer mounted on the test section is chosen as y. The

system shown in Figure 3-11 can be expressed in frequency domain as

Y = H1X, + HX2 +N (3.2)

where X1, X2, N and Y are the Fourier transforms of x, X2, n andy respectively.

The relevant power and cross spectra are given by


G (= E[XI-.X2]

2
Gly(f)=T E [XI Y]

G,(f) = TE[X -Y]

G, (f) = 2 E[ Y]
T

where denotes the complex conjugate.

The ordinary coherence functions are calculated as follows:



G -G^
2
S G I- ) (3.4)
2
2 G(f)G G2
G 22 -Gyy

Assuming A22 (f) e (0,1), the noise spectrum G,, is computed as


G, = G H1 G11 + H;H 2G12 + H2H1G(21 +-I H2 G22
(3.5)
Gnn = G, G6








36

where the frequency response functions H1 and H2 are calculated according to the

formula given below:


G,,G,, -G,,G
1 2
H GG22G G12G2y
(3.6)

G-zG2 -G---G
Gl G2y G21Gy



The cross terms involving H1 and H2,, i.e. H1H2G12 and H2H1G21, arise because of

the interaction between xi(t) and x2(t).

The multiple coherence function, that represents the correlated output power in the

test section acceleration due to the input fan and pump acceleration, is given by


7x(f) = (3.7)


Results and Conclusions

Results of the vibration data for the case when the air speed is 64.5 m/s and water

speed is 0.5 m/s are presented. Results for all other test cases, tabulated in Table 3-1, are

presented in Appendix B.

The ordinary coherent output spectra, 72yGy and y'yG are shown in Figure 3-13.

Also, the noise contributions G,, and the multiple coherent output spectrum (i.e., the

combined vibration contributions of the fan and pump) G,, are compared to the output

spectrum Gy in Figure 3-13. For comparison, the measured noise floor spectrum,

obtained when the blower and pump are off, is also plotted.










The ordinary coherence functions between the inputs, 72, and between each input

and the output, ~, and r2 are shown in Figure 3-14. Also, the multiple coherence

function, y2x, is shown in Figure 3-14.

The results indicate a large content of the output spectrum is a part of the "noise" Gnn

that is not modeled. However, the multiple coherence function (and other ordinary

coherence functions) is close to 1 at certain frequencies indicating that contribution at

these frequencies are largely due to the stated inputs x1 and x2, i.e. the z axis component

of the vibration measured by the accelerometers near the fan and pump, respectively.
































































10o
0


100 200 300 400 500 600 700 800
1 100 200 300 400 500 600 700 800


10-5



10
~6-10







10-



- 1 0







10'


w w


0 100 200 300 400 500 600 700 800


IG
nn
SG noise floor
100 200 300 400 500 600 700 800
Frequency (Hz)


Figure 3-13 Top two plots show the ordinary coherent output spectrum due to the inputs.
The bottom two plots compare the noise spectrum with the output spectrum, the multiple
coherent output spectrum and the noise floor.


100 200 300 400 500 600 700


h
~
















1 --1 ------
0.8---

0.6 -- -- -- -
0.6


0.4
0,2
0 100 200


300 400 500 600 700 800





300 400 500 600 700 800


0.8-
0.6 -

0.4 -
0.2


2l


- ------- ---


0 100 200 300 400 500 600 700 800

1
08 100 20 -0 -- -- 600 -- 800-----------

0 .6- - _ _-__ _
0.4 ---- --
06


0 100 200 300 400 500 600 700 800


0

0


0


1


.6- ----- --------- ------- ---





0 100 200 300 400 500 600 700 800
Frequency (Hz)


Figure 3-14 The topmost plot shows 71, the coherence between the inputs. The next

two plots are y and r2 the ordinary coherence functions between each input and the

output. The bottommost plot is yx, the multiple coherence function.


To distinguish the effect of x, from x2 on the output, the characteristics of the blower


and pump are studied. For the blower and the pump, the blade passage frequencies of the


IS\i








40

fan and the pump impeller are identified in a separate experiment using a BK condenser

microphone.

The blade passage frequencies of the fan and pump impeller are determined for the

conditions listed in Table 3-1. To determine the blade passage frequency of the fan, the

pump is switched off and acoustic pressure measurements are obtained very close to the

entrance of the blower with the microphone. A similar exercise is repeated for the pump

impeller. The specifications of the microphone are given in below. The results are

compared to Gyy in Figure 3-15.

A strong tone at 364 Hz is noticed in the spectrum of the microphone data near the

blower. A second tone of lesser intensity is noticed at 728 Hz in this spectrum. The

second tone is inferred as the harmonic of the 364 Hz tone. In addition, a range of

smaller tones is noticed at 40 Hz, 81 Hz, 121 Hz, 162 Hz and 202 Hz. This is explained

as follows.

The fan in the blower is known to have nine blades. The blade passage frequency is

calculated as the number of blades times the frequency of revolution of the fan. The

result of dividing the frequency of the first tone, i.e. 364, by the number of blades is very

close to 40 Hz. Therefore, there is a high probability that the first peak of the range is a

result of eccentricity in the mounting of the fan because it is likely to have the effect of

producing tones at the frequency of revolution. The subsequent peaks at multiples of 40

Hz are most likely to be the harmonics of the tone at 40 Hz.

The output spectrum Gyy has a strong tone at 364 Hz and 728 Hz, which is mostly due

to the strong vibration exhibited by the blower at the blade passage frequency and its first

harmonic. Also, discernible peaks are noticed at 40 Hz, 81 Hz and 202 Hz that arise









41


mostly as an effect of eccentricity in the mounting of the fan in the blower section. In


addition, several peaks are noticed spaced approximately 40 Hz apart after 364 Hz.


-G
Gyy
SFAN MIC SPECTRUM
FAN MIC NOISE SPECTRUM


I I
I I Ih/'^ ^


J- i


10-2


10-4

10-6 '
10-

10'

10-
10

10-12

10-14
0


500 600 700 800


-G
G y y
PUMP MIC SPECTRUM
i -P -- M PUMP MIC NOISE SPECTRUM



It~lrl 1 CI I
- I Iil ^J jk d ', ,


100 200 300 400
Frequency (Hz)


500 600 700 800


Figure 3-15 Comparison of output spectrum with BK microphone measurements. In the
top plot, the microphone is placed very close to the blower, with the pump off. In the
bottom plot, the microphone is placed very close to the pump, with the blower off


For the case when the air speed is 60 m/s and water speed is 0.5 m/s, the comparison


of the microphone spectrum near the pump with the output vibration spectrum Gy is


100 200 300 400
Frequency (Hz)


10-2

10-4


10-


10-8

10
10 -


10-12


10-14
10
0









42


inconclusive. This is because vibration levels induced by the pump are very small


compared to the fan. The ordinary coherence function ( 2) between the inputs, plotted


in Figure 3-14, shows that the pump experiences significant vibration from the fan.


To isolate the influence of the pump vibrations on the on the test section, the vibration


data for the case when the fan is switched off is analyzed. The water tunnel speed is set


at approximately 0.5 m/s. Along with the vibration measurements, BK microphone


measurements are made. The results are shown in Figure 3-16.



10-2
G
YY
-PUMP MIC SPECTRUM
-PUMP MIC NOISE SPECTRUM
-o4
104 --- -- --- --- --- --- --- --- -- ------ ------------ --



10-6


I I I I
101-80I I,




10-12 -- --

-o ----- --------- ____- -__-__-_-_-_-_-_-_-_-_-_-_--



10-14 L L L
0 100 200 300 400 500 600 700 800
Frequency (Hz)



Figure 3-16 Comparison of the output spectrum with microphone measurements for the
case when the fan is off and the water tunnel speed is 0.5 m/s.

The output spectrum is almost three orders of magnitude less than that of the previous


case analyzed. The results show that harmonics as well as sub harmonics are present in


both the output spectrum and the microphone data.








43

In conclusion, the fan vibrations are much greater than the pump. The test section

vibration after removing the fan and pump contributions is still much greater the

measured than noise floor, indicating other significant inputs (e.g., the other vibration fan

and pump components).















CHAPTER 4
INTERFACE EDUCATION USING LIGHT SHEET

This chapter describes a simple method used to characterize the air water interface of

the facility under different combinations of air and water speeds. The results are helpful

in determining the amount of fluctuations in the air-water surface at a particular

condition. A measure of both root mean squared amplitude and slope are obtained using

a high-resolution camera in conjunction with a pulsed laser light sheet. A simple edge-

detection algorithm is employed to determine the location of the air-water interface. The

aim is to educe quantitative information of the air-water interface.

Experimental Set-up

The experimental set-up consists of a pulsed Nd:YAG laser and a digital camera (TSI

model PIVCAM), as shown in Figure 4-1. The Nd:YAG and associated optics generate a

thin light sheet (-lmm thick) required to illuminate the air-water interface. The camera

resolution is 1016 by 1000 pixels and acquires image pairs that are synchronized with the

laser. To enhance the scattering at the air-water interface, a fluorescent dye Eosine Y is

used. Some of the critical parameters used with the laser and camera are listed in Table

4-1. A span of little less than 4.5 in. is covered (corresponding to the range of 5.5 in.-10

in. from the splitter plate), yielding a spatial resolution of 209 pixels/in, and 223 pixels/in.

in horizontal and vertical world coordinates, respectively.

As shown in the camera is Figure 4-1, the camera is positioned to obtain a perspective

view of the air-water surface. A perspective view avoids an overlap between the desired








45

mid-plane edge and the unwanted edge at the sidewall. This simplifies the air-water

interface education algorithm.


Figure 4-1 Experimental set-up with Nd:YAG LASER and PIVCAM

Table 4-1 Critical parameters of the laser and camera


Procedure

A sequence of images at the conditions shown is acquired for the conditions listed in

Table 4-2. The numbers in the table correspond to the speed of the air/water streams. A

total of 200 images are taken at the pulse repetition rate (PRR) specified in Table 4-1.


Pulse separation (At) 500 ps

Camera shutter open time 20 ps

Pulse rep rate (PRR) of laser 15 Hz (max)

Pixel array size 1016 x 1000













Table 4-2 Test matrix for interface education

Air (Hz|ms1)
010.0 313.1 515.2
Water (Hzlms )
Pump off
0|0.0
0.1
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9


An edge detection algorithm is applied to detect the air-water interface from the

images. The edge detection algorithm used is based on the sobel method [Gonzalez and

Woods 2001]. An example for the case with an air speed 5.2 m/s and a water speed 0.4

m/s is shown in Figure 4-2. In all images, the flow direction is from left to right.













200 200

.0L400 400 -

>600 > 600

800 800
1000.I 1000

200 400 600 800 1000 200 400 600 800 1000
X pixels X pixels


Figure 4-2 Sample original image (left) and corresponding output from edge detection
algorithm (right). The flow direction is from left to right as indicated.

Since the edge detection algorithm detects all edges, several extraneous edges can be

identified as shown in Figure 4-2. The fact that sufficient scattering occurs at the air-

water interface ensures that the interface is always a nearly continuous edge. This fact is

exploited to prune the image of the unwanted edges or features.

Interface Detection

The key idea is to obtain the interface image as a binary image (i.e., the interface

pixel has a value of 1 and 0 otherwise). The output of the edge detection algorithm can

be pruned to provide the interface image shown in Figure 4-3. Since the interface is

detected as a clear continuous edge, the occurrence of the first pixel value of 1 found in a

column search of the edge image is assumed to be the interface.











300
Mid-plane interface Mid-plane interface
350 200

400 400 --

450 >-600

500 800
Side-wall edge
550 1 1000
0 200 400 600 800 1000 200 400 600 800 1000
X pixels X pixels


Figure 4-3 A sample edge image and the corresponding interface image as a result of
pruning.

The interface images of a whole sequence of 200 are then averaged, thereby obtaining

an average interface for a particular tunnel condition. Also, the standard deviation value

of the fluctuations of the interface location is calculated as


1000 1000 ,2
I 0(KOK) IJ OK
1000 1000 (4.1)
I IY
J=1 J=1


where Ii is the a column vector of Y pixel values atjh X pixel, K =[1,2,...,1016],


denotes element by element multiplication and L ] denotes rounding off to the nearest


integer value. The advantage of computing the standard deviation value is two fold.

First, it gives a measure of the fluctuations of the air-water interface. Second, it is

essential to identify outlier points. An average interface image and the corresponding

standard deviations (from mean at an average interface image pixel coordinate) are

shown in Figure 4-4.
















200


,400- -
. 0
>600


800


1000
200 400 600 800 1000
X pixels


30

S25

20
o
S15
o
10,


C 5


0 200 400 600 800 1000
X pixels


Figure 4-4 Average interface and standard deviation of air-water interface. Number of
statistics: 200


Outlier Rejection


The interface detection algorithm reveals outliers as shown in Figure 4-5. The


presence of outlier points in the edge image results in high standard deviation values at


certain X pixels as shown in Figure 4-4. The outlier points and the interface pixel


distribution for a column of sum of sequence of interface images are shown in Figure 4-5.


30


'-- Outlier points

-- -


200 400 600 800 1000
X pixels


0)
025
0)
20
0
2 15

10
o 'utiiher
t- point


C0 .-.I,........- ,, ... . l... ,
0 200 400 600 800 1000
Y pixels


Figure 4-5 Superposition of sequence of edge images before outlier rejection and a
column (at 188th X pixel) of sum of sequence of interface images with an outlier point.


200


~400
. x
>600


800


1000








50

The outlier criteria used here is based on comparing standard deviation. First, the

mean p and standard deviation a of the data set are calculated for each column in an


image. Then, all data points that lie outside the interval (p 3u, p + 3) are disregarded


according to Modified Tau Thompson criterion [Holman 2000]. Applying this to the sum

of sequence of interface images, the result, shown in Figure 4-6, is obtained.


30

0 25
200 2

S20
400
U)
.x 4 15
600 "-
_10
800

1000 00 .-.. .... ,* ...... .... .. .-.. ........
200 400 600 800 1000 0 200 400 600 800 1000
X pixels Y pixels


Figure 4-6 Sum of sequence of edge images after outlier rejection and a column (at 188th
X pixel) of sum of sequence of interface images without any outlier points.

After the outliers are eliminated, the interface image and standard deviation values

are calculated again. The outlier criterion does not significantly affect the mean of the

data set; hence the average interface image is almost unchanged (Figure 4-7). Analysis

of the two images in Figure 4-7 shows that the difference between the interfaces is not

more than one pixel.









51





200 200


u0400 -400 -
x x
0- a
>-600 >-600


800 800


1000 1000
200 400 600 800 1000 200 400 600 800 1000
X pixels X pixels


Figure 4-7 Comparison of interface images without and with outlier rejection. The
average images are shown before outlier rejection (left) and after outlier rejection.

However, the application of outlier criterion has significant effect on the standard

deviation of the interface image. The standard deviation values before and after outlier

rejection are shown in Figure 4-8.


30 30

25 8 25

>20 20

15 5 15

1 1
(D (





0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
X pixels X pixels


Figure 4-8 Comparison of standard deviation values before and after outlier rejection.

Transformation and Calibration

Since the images are taken with a camera, the original spatial signal (image)

undergoes a perspective transformation. Assuming the camera is a point source, a








52

reverse projection transformation is required to map the image to world coordinates so

that the interface location is determined in physical dimensions.

For this purpose, a calibration image is obtained with the camera. In the calibration

image, the world coordinates of specific points in the image are known. In this case, the

calibration image is a series of squares whose vertices are known in world coordinates.

The lines of the square that are captured by the camera are at least 5 pixels thick.

Projection transformation (camera view) of these squares will transform parallel lines

into lines that intersect at some point within or outside the image. The "n" number of

intersection points defines the projection transformation as an "n" point projection. Since

the image is of a plane as a pixel map in this case, the projection is a two-point

projection. The camera view (transformed) and the reverse projection transformation of

the calibration image are shown in Figure 4-9. The transformation is achieved by means

of trial and error by ensuring the quadrilaterals are mapped to rectangles whose vertices

have a pixel difference of less than 1.


150 150
(0) C0)
.250 x250
> >

350 350
300 400 500 600 700 800 300 400 500 600 700 800
X pixels X pixels


Figure 4-9 Projection transformation and reverse projection transformation of the
calibration image.

The following transformation matrix achieves the desired result:












0.9322 -0.0273 -0.0273
T = 0.0470 0.9517 -0.0405 (4.2)
-0.0470 0.0078 1.000


xy x y
0 0 -0.047 0.0078
T- 1 0 = 0.91 -0.02 (4.3)
01 1 0
1 1 1 1.001


where (x, y) are the world coordinates and (x, y) are image coordinates.

Calibration of the reverse projection transformation image enables measuring the

interface in physical dimensions (inches) instead of pixels. Knowing that the squares in

the calibration image are of length 1 in., the following results:


World X axis: 209 pixels = 1 in.
(4.4)
World Y axis: 223 pixels = 1 in.

Now, applying the reverse transformation to the average interface image, we get the

average interface in the world coordinate system as shown in Figure 4-10.



-2
200
-1
,400 ---
x o0
>-600

800
2
1000
200 400 600 800 1000 5 6 7. i 8 9 10
X pixels X inches


Figure 4-10 The average interface image and the corresponding reverse transformation to
world coordinates.












Interface Slope Calculation

The first step in calculating the slope of the interface is fitting a polynomial of an


appropriate order. A visual inspection of the interface image (Figure 4-11) shows that

there are four extremums. Therefore, a polynomial fit of 5th order is chosen as shown in


Figure 4-11. This order is chosen to ensure minimal error in the polynomial fit while


maintaining a smooth slope. The value of maximum error for possible polynomial fits is


tabulated in Table 4-3.


Table 4-3 The maximum error between the data and polynomial fit.


Polynomial fit order Maximum error (pixels)
3 5.1

4 3.8

5 1.4

6 1.5
7 1.8
8 1.5


5 6 7in e8
X inches


-0.9

-0.8

-0.7
U)
_-0.6
C
-0.5

-0.4


9 10


5 6 7in 8
X inches


Figure 4-11 Interface image in world coordinates and corresponding polynomial fit.

With the polynomial fit known, the slope of the interface can be calculated by


differentiating the expression. The corresponding slope is also shown in Figure 4-12.


-0.9

-0.8

-0.7
U)
-0.6

-0.5

-0.4

-0.3


Interface
polyfit


9 10


I_ ---- I---- ----- -








55

Comparison of every case with the baseline case of zero velocity of air and water tunnel

will give the relative interface displacements.

x 10-3

-2 Interface
polyfit


5 6 7 8
Xinches


9 10


5 6 7 8
X inches


Figure 4-12 Polynomial fit, of order 5, of average interface image (left) and
corresponding slope of interface (right).

The next chapter will summarize with conclusions and also list the possible future

endeavors related to the facility.


9 10















CHAPTER 5
CONCLUSIONS AND FUTURE WORK

The University of Florida air-water shear-layer facility is unique for testing

algorithms and characterizing MEMS sensors related to supercavitating vehicles. The

facility allows for sensor characterization and algorithm testing by emulating

supercavitating conditions satisfactorily.

Velocity profiles measured for the air stream and the water stream are found to be

uniform. The mean flow quality is very good in both the sections. Future work can be

done for more accurate measurements of the velocity profile, variations in speed as well

as angular using more precise measurement tools such as particle image velocimetry and

LASER Doppler velocimetry. Also, influence of fan and pump vibration on the

fluctuating flow quantities has to be assessed.

Operation of the air tunnel at high speeds, with free surface condition, entrains

significant amount of water from the water tunnel thereby rapidly reducing the water

level in the water tunnel. As a remedy to this problem, a feedback-based system should

be employed to maintain desired water levels in the water tunnel. Such a feedback

system would comprise of a variable throughput pump, a water level sensor with very

low reaction time and a feedback circuit.




















APPENDIX
TUNNEL OPERATION PROCEDURE


l-_i.z


File Edit Operate Tools Browse Wndow Help


'ilc patlh I l: .- r.r
Jfl Air speed control Heise Reading Xducer reading I'i,. i...
r .ll LU l3 l I r'rsion ... ,-I.
Channel -in- IR j- -







AutI[IAQ d sol on 1

.... .
4- .. ....... ..








I I.
Range


1. Set Air speed control to "auto."


2. Leave the channel as channel "0."


3. Choose the voltage increments and ceiling to be used for the experiment, with a


maximum ceiling of 10 Volts.


4. Set the file path to which data will be written.


5. Choose the appropriate IP address for the traverse to be used.









58

6. On the S7 Operator Interface panel on the side of the air tunnel, press the "MON"

button.

7. Hit "ENT" to enter the "Basic Parameters Group."

8. Press the up arrow four times, to access the "Frequency select" menu. Press "ENT."

9. Press the up arrow to bring the displayed setting to "Terminal." Press "ENT." This

brings the tunnel under DAQ control.

10. Press "MON" twice to return to the main display.

11. Press the "RUN" button on the panel.

12. Run the VI.

13. When the experiment is complete, press the "STOP" button on the panel.


Eile Edit Operate Tools Browse Window Help
0 _6 l. ll' .F ___D1tI-
hie path l Ji-' r --;. I
jl Air speed control Heise Reading Xducer reading H.
Mdlmald Conltrol VISA season
Channel I 20 '.. "-e -',] 1
hnannl l --2



-l I I

Au.to DAO LI




Output A rla










1. Set Air speed control to "auto."

2. Set the Channel to channel "1," to enable water speed control instead of air.

3. Choose the voltage increments and ceiling to be used for the experiment, with a

maximum ceiling of 8 Volts.

4. Set the file path to which data will be written.

5. Choose the appropriate IP address for the traverse to be used.

6. On the S7 Operator Interface panel on the side of the water tunnel, press the "MON"

button.

7. Hit "ENT" to enter the "Basic Parameters Group."

8. Press the up arrow four times, to access the "Frequency select" menu. Press "ENT."

9. Press the up arrow to bring the displayed setting to "Terminal." Press "ENT." This

brings the tunnel under DAQ control.

10. Press "MON" twice to return to the main display.

11. Press the "RUN" button on the panel.

12. Run the VI.

13. When the experiment is complete, press the "STOP" button on the panel.
















REFERENCES


Banazynski, K. A. (2000). Air/Water Shear Layer Facility Manual, Engineering
Laboratory Design Inc., St. Louis, MN.

Bendat, J. S., Piersol, A. G. (2000). Random Data: Analysis and Measurement
procedures, Wiley-Interscience, New York.

Billet, M., Tip vortex cavitation, http://www.arl.psu.edu/areas/cavitation/cavitation.html.
Accessed on 06/10/2002.

Chandrasekaran, V., Chow, E. M., Kenny, T. W., Nishida, T., Sheplak, M. (2001a).
"Through Wafer Electrical Interconnects for MEMS Sensor," Proceedings of the
American Society of Mechanical Engineers. New York. NY. pp 232-245

Chandrasekaran, V., Li, X., Nishida, T., Cattafesta, L. N., Li, J., Sheplak, M. (2001b).
"Thermoelastically Actuated MEMS Ultrasonic Transducer," Presented at 142nd
Meeting of the Acoustical Society of America. Ft. Lauderdale, FL.

Chandrasekaran, V., Chow, E. M., Kenny, T. W., Nishida, T., Cattafesta, L. N., Sankar
B.V., Sheplak, M. (2002). "Thermoelastically Actuated Acoustic Proximity Sensor With
Integrated Through-Wafer Interconnects," Presented at Solid-State Sensor and Actuator
Workshop. Hilton Head. SC.

Gonzalez, R. C. and Woods, R. E. (2001). Digital Image Processing, 2nd edition,
Addison-Wesley, New York.

Holman, J. P. (2000). Experimental Methods for Engineers, McGraw Hill Higher
Education, New York.

Kulkarni, S. S., Pratap, R. (1999). "Studies on the Dynamics of a Supercavitating
Projectile," Applied Mathematical Modeling 24: 113-129.

Kunz, R. F., Lindau, J. W., Billet, M. L., Stinebring, D. R. (2000). "Multiphase CFD
Modelling of Developed and Supercavitating Flows," Applied Research Laboratory,
Pennsylvania State University.

Li, X., Lasson, E., Sheplak, M., Li, J. (2002a). "Phase-Shift-Based Time Delay
Estimators for Proximity Acoustic Sensors," IEEE Journal of Oceanic Engineering. 27:
No. 1. 47-56.











Li, X., Wu, R., Sheplak, M., Li, J. (2002b). "Multifrequency CW-Based Time-Delay
Estimation for Proximity Ultrasonic Sensors," To appear in IEE Proceedings F: Radar,
Sonar, and Navigation.

Li, X., Wu, R., Rasmi, S., Li, J., Cattafesta, L., Sheplak, M. (2002c). "An Acoustic
Proximity Ranging System for Monitoring the Cavity Thickness," Submitted to IEEE
Transactions on Ultrasonics, Ferroelectrics, and Frequency Control.

Rightley, P. M, Lasheras, J. C. (2000). "Bubble Dispersion and Interphase Coupling in a
Free-Shear Flow," Journal of Fluid Mechanics 412: 21-59.

Sheplak, M., Cattafesta, L. N., Shyy, W., Kurdila, A. J., Nishida, T. (1999). Advanced
Technology Development for the Control of High-Speed Supercavitating Vehicles,
Interdisciplinary Microsystems Group, Mechanical and Aerospace Department,
University of Florida.

Stinebring, D. R., Cook, R. R., Dzielski, J. E., Kimerer, N. B., Kunz, R. F., Miller, T. F.
(2000). High-Speed Supercavitating Vehicles, Applied Research Laboratory,
Pennsylvania State University.

Stinebring, D. R., Billet, M. L., Lindau, J. M, Kunz, R. F. (2001). Developed Cavitation-
Cavity Dynamics, Applied Research Laboratory, Pennsylvania State University.

Stutz, B.,. Reboud, J. L. (1997). "Experiments on Unsteady Cavitation," Experiments in
Fluids 22: 191-198.

Tulin, M. P. (1961). Supercavitating Flows, Handbook of Fluid Dynamics, V. L. Streeter,
New York, McGraw Hill: 12.25 12.46.

Werner, D. J. (1998). Technology Assessment of Hydrodynamic/Supercavitating
Technologies, Alliant Techsystems Inc.
















BIOGRAPHICAL SKETCH

Srihari Rasmi received his Bachelor of Technology degree from the Indian

Institute of Technology, Madras, India, in 1999 in naval architecture. During December

1998 to April 1999, he received a scholarship to pursue his research interests related to

stability of ships at Hochshule Bremen in Bremen, Germany. From August-December of

1999, he was an academic visitor at the Ship Stability Research Center at the University

of Strathclyde in Scotland. Since January 2000, he has been pursuing his M.S degree in

the Department of Aerospace Engineering, Mechanics, and Engineering Science at the

University of Florida in the areas of experimental fluid dynamics and signal processing.