UFDC Home  myUFDC Home  Help 



Full Text  
REMOTE DETECTION OF HYDROGEN LEAKS USING LASER INDUCED RAYLEIGH/MIE SCATTERING By SAMEER PARANJPE 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 2004 Copyright 2004 by Sameer Subhash Paranjpe ACKNOWLEDGMENTS The author would like to thank Dr. Jill Peterson for her guidance and support. I would also like to thank my fellow students Raghuram Vempati, Philip Jackson, Murray Fisher, Matthew Gabriel and Ryan Ferguson for their assistance in various portions of the project. TABLE OF CONTENTS page A C K N O W L E D G M E N T S ......... .................................................................................... iii LIST OF TABLES ............. ............................... .......... .......... .. vi LIST OF FIGURE S ......... ..................................... ........... vii ABSTRACT .............. .................. .......... .............. xi CHAPTER 1 INTRODUCTION ............... ................. ........... ................. ... .... 1 2 LITER A TU R E REV IEW ............................................................. ....................... 6 M ie Scattering .................................................................................................... ...... 6 R ayleigh Scattering ........... ..................................................................... ....... .. . Buoyant Jet Theory ............... ................ ......... ......................... .9 3 THEORETICAL FRAM EW ORK .................................... ......................... .. ......... 10 Elastic and Inelastic Light Scattering .................................. .................................... 10 Elastic Scattering M echanism .......................................................... ............... 10 M ie T theory ................................... ............................11 Calculation of intensity distribution functions I1 and I2 ...................................14 R ayleigh Theory .................................................. ..... .................... .. ... 15 Photon arrival rate calculations for Rayleigh and Mie theory.............................17 Scattering Cross Section Considerations.........................................................18 Buoyant Jet Theory ................. ............................ ..... ...... 18 Froude N um ber C alculation .............. ................................. ............... ... 19 Buoyant Jet Profiles and Concentration Variation ...........................................20 4 EXPERIM ENTAL SCHEM E .............................................................................23 Collection at 900 and B ackscatter......................................... .......................... 23 B ack Scatter C onsiderations..................................................... ................... .30 D ata R recording Procedure.............................. ........................... ... .............30 5 RESULTS AND DISCU SSION ........................................... .......................... 32 Comparison of Theoretical and Experimental Photon Arrival Rate...........................32 Calculation of Theoretical Rayleigh Photon Arrival Rate. ................................32 Calculation of Theoretical Mie Photon Arrival Rate .......................................33 Calculation of Total Theoretical Photon Arrival Rate. .....................................38 Calculation of Experimental Photon Arrival Rate................... ..... ............38 D ata A analysis Techniques ................................................ .............................. 41 Integrated area m ethod. .............................................. ............................. 42 Peak V oltage M ethod. ............................................... .............................. 43 Analysis of Recorded Data .......... ........ ...................................... 46 C ase I: A rgonion laser............ .................................................... ........ 46 P eak voltage variation ............................................ ........... ............... 46 N orm alized peak voltage variation ................................... .................48 Standard deviation profiles...................................... ......................... 48 Case II: Pulsed Nd:YAG laser; pure helium ............................................... 50 P eak voltage variation .............................................. ......... ............... 50 N orm alized Peak Voltage Variation .................................... ............... 51 Standard deviation profiles...................... ........................... 51 Case III: Pulsed Nd:YAG laser; 20%helium ,80% nitrogen.............................52 P eak voltage variation ............................................ ........... ............... 52 Norm alized peak voltage variation. .................................. .................53 Standard D aviation Profiles. ............................................. ............... 54 M easurements in Backscatter ................................. ................................... 55 6 CON CLU SION S .................................... ... ........... .......... ........... 58 LIST OF REFEREN CES ................................................................... ............... 60 B IO G R A PH IC A L SK E TCH ..................................................................... ..................62 v LIST OF TABLES Table page 31 Scattering cross section at a wavelength of 532 nm............................................18 32 Froude number calculations show that the criteria for buoyant jet is met for all combinations of flow fluids, nozzle diameters and downstream distances.............20 51 Particle counter data show the particle distribution in the lab ..............................34 52 Shows that average peak voltage recorded after 300 pulses in shear layer is a conv erg ed m ean .................................................. ................. 4 5 53 Control volume to leak diameter ratio for all three cases shows that the ratio is far less for backscatter than 900 scattering .................................... ............... 57 LIST OF FIGURES Figure page 31 M ie scattering geom etry ......... ......... ................ ........................ ............... 11 32 Ii (perpendicular) and I2 (parallel) intensity distribution functions from the interactive w ebpage .................. ................................. ..... .. ........ .... 15 33 Ii (perpendicular) and I2 (parallel) intensity distribution functions from McCartney. 15 34 Instantaneous and time averaged profiles of a typical buoyant jet...........................21 35 Concentration variation with r for z = 2 ....................................... ............... 22 41 Experimental scheme for collection of scattered light at 90.............. ................24 42 Experimental scheme for collection of scattered light at a 1800 (back scatter) .......25 43 Nozzle mounting showing three dimensional motion capability...........................27 43 Photomultiplier tube linearity tested as a function of flash lamp voltage ..............29 44 Calculation of nozzle traverse distances ....................................... ..................... 31 51 Theoretical photon arrival rate calculations. ...................................................33 52 Model M shows maritime distribution of aerosols (McCartney 1979, page 139) ...35 53 M odel M duplicated on a lognormal scale................................... ............... 36 54 Comparison of number density of maritime and lab aerosols..............................37 55 Effect of maritime and lab aerosol distribution on Mie scattered intensity ............37 56 Typical waveform of a burst seen on the oscilloscope....................................39 57 Points in shear layer where measurements are made ............................................40 58 Comparison of experimental and theoretical photon arrival rates .........................41 59 Convergence studies of normalized area and averaged area...............................42 510 Waveform with varying glare at four downstream locations ...............................43 511 All four waveforms normalized with their individual peaks.................................44 512 Voltage variation for ArgonIon laser. ............................... ............................... 47 513 Normalized peak voltage variation for argonion laser.........................................48 514 Percent standard deviation variation for near field case. .......................................49 515 Percent standard deviation for far field case .................................... .................49 516 Voltage variation for pulsed Nd:YAG laser.................................. ............... 50 517 Normalized peak voltage variation for Nd:YAG laser.................. ... ............ 51 518 Standard deviation variation for Nd:YAG laser for pure helium...........................52 519 Relative size of beam and leak diameter ........................................ .............52 520 Voltage variation for the pulsed Nd:YAG laser for the mixture.............................53 521 Normalized peak voltage variation for Nd:YAG laser for the mixture .................54 522 Percent standard deviation variation for Nd:YAG laser for the mixture of helium and nitrogen to simulate the optical properties of nitrogen................. ......... 54 523 Schematic ofbackscatter on a time basis shows separation between beam dump glare and scattered signal from the nozzle. ................................... ............... 55 524 Normalized area variation shows a reduction in scattered intensity in presence of helium (backscatter); Nozzle diameter = 1/4" ; Re =500; Fr = 290 .........................56 NOMENCLATURE a = particle radius (m) C = concentration of the flow fluid (kg/m3) C* = dimensionless density CPMT = photomultiplier tube calibration constant D = diameter of the nozzle (m ; ") E' = Electric vector of the incident wave (V/m) Fr = Froude number =Re/Gr2 G= gravitational acceleration (m/s2) Gr = Grashof number = g(papo)D3/po" H' = Magnetic vector of the incident wave (N/Amperem) o = Incident laser power (W; photons/pulse) 1 = length of control volume (m) m = mass flow rate (kg/s) n = gas index of refraction at known reference conditions n(r) = number density per radius interval (molecules/m3/micron) N= molecular number density (molecules/m3) Nd = number of data points. r = radial distance from jet centerline (m) rxy= cross corelation coefficient Re = Reynolds number = pvD/ u = 4m/ 7tDg v = velocity of buoyant jet (m/s) V= photomultiplier tube voltage (V) x = percent helium y = distance from jet centerline normal to beam z = downstream distance (mm) Greek Symbols a = size parameter = 27na/X P = spread angle 0 = angle of observation measured from the forward to scattering directions. p = scattering angle y = wave function rI = optical efficiency of transmitting and collecting lenses Q = solid angle of the collection optics X = wavelength of laser light (nm) c = differential scattering cross section (m2/sr) p = density of gas (kg/m3) [ = dynamic viscosity of gas (Pas) u = dynamic viscosity of gas Subscripts a = ambient cl = centerline i = species m = mixture o =jet exit condition 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 REMOTE DETECTION OF HYDROGEN LEAKS USING LASER INDUCED RAYLEIGH/MIE SCATTERING By Sameer Paranjpe December 2004 Chair: Jill Peterson Major Department: Mechanical and Aerospace Engineering The current study examines the use of laser induced Rayleigh/Mie scattering as a means of remotely detecting hydrogen leaks. An axisymmetric vertical buoyant jet at a Reynolds number of 500 was used to simulate the hydrogen leak and the scattered signal indicating hydrogen concentration was examined at different downstream locations. Helium was used as a substitute for hydrogen for safety reasons. The scattering cross section of hydrogen is 0.23 times the scattering cross section of air and the scattering cross section of helium is 0.015 times the scattering cross section of air. A mixture of 20% helium and 80% nitrogen was also used at the same Reynolds number of 500, since the scattering cross section of this mixture equals the scattering cross section of hydrogen. The principal challenges in RLS detection were electronic shot noise and Mie scattering. The electronic shot noise was found to induce less than 0.1% uncertainty for an averaging time of 1 second. A Nd:YAG pulse laser operating at a wavelength of 532 nm was used and the scattered signal from the helium leak was collected at 90 to the incident beam and focused onto a photomultiplier tube. The signal from the photomultiplier tube was read through a high speed digital oscilloscope. The repeatability and reproducible of the data was established using a set of convergence studies. It was found that the amplitude of the Rayleigh/Mie signal decreased by 25% and the standard deviation increased by 45% in the presence of helium. This increase in standard deviation is more than the established values of 305. Non dimensionalized profiles collapsed to a classic similar shape, further documenting experimental results. CHAPTER 1 INTRODUCTION Hydrogen is an attractive fuel source. However, leak detection is essential if it is to become a widespread, easily used and safe source of energy. It is relatively simple to determine whether a system is leaking hydrogen by identifying pressure drops. Finding the source of the leak however can be time consuming, costly and dangerous. The conventional method of leak detection involves the use of contact sensors. The following are the typical varieties of contact sensors commonly used for leak detection. Catalytic bead sensors. These sensors consist of two beads surrounding a wire operating at a temperature of around 4500C. One of the beads is passivated. This ensures that it does not react with gas molecules. The other bead is coated with a catalyst to promote a reaction with the gas. The beads are generally placed on separate legs of a Wheatstone bridge circuit. When hydrogen is present, there is no measurable effect on the passivated bead, but there is a significant effect on the catalyzed bead. The increase in heat increases the resistance in that leg of the Wheatstone bridge circuit, which in turn changes the bridge balance signal, and this serves as the sensor signal. These sensors are usually used in the 1 to 5 % hydrogen range. The response time of the sensor varies, ranging from 10 to 30 seconds for fullscale response. Semiconductor sensors. These sensors use semiconducting oxides whose electrical resistance changes in the presence of hydrogen due to a reduction reaction. These sensors generally operate at temperatures above ambient. The disadvantage of these sensors is that the oxides change their resistance as the oxygen concentration in the environment changes, making these sensors unsuitable for such environments. They have a fast response time and detection limits of 01000 ppm hydrogen. Electrochemical sensors. Electrochemical sensors are composed of an anode and cathode sandwiching a chemically sensitive electrolyte. When hydrogen passes over the electrolyte, a reversible chemical reaction occurs. This generates a current proportional to the gas concentration. However, oxygen is required to ensure chemical reversibility. This implies that the sensor is not environmentally independent. Electrochemical sensors are typically used to detect hydrogen in the range of 100 to 1000 ppm. Response times can be as low as several seconds, although typically these sensors are specified at 30 to 50 seconds for fullscale response. Resistive palladium alloy sensors. The surface of palladium acts catalytically to break the HH bond in diatomic hydrogen and allows the monatomic hydrogen to diffuse into the material. Palladium can dissolve more than 600 times its volume in hydrogen. The level of dissolved hydrogen proportionally changes the electrical resistivity of the metal. No other gases or environmental controls are necessary for these measurements. Hydrogen field effect transistor. By using palladium as the gate material for a standard field effect transistor, small changes in the resistivity of the palladium produce large changes in the currentvoltage characteristics of the FET. This sensor technology works well in the range of 50 to 1000 ppm range of hydrogen. Although all these systems can effectively detect hydrogen leaks, the sensors are intrusive and have to be physically inserted into the suspect area. However due to the low combustion limits (4%) of hydrogen, for safety issues in most applications, it would be advantageous if the leak could be detected without actually inserting a probe in the suspected area. The current study explores a novel remote detection technique. Laser Induced Rayleigh/Mie Scattering This method uses a laser directed at a suspected leak source. As the laser pulse moves through air, the electromagnetic wave interacts with aerosols in the atmosphere and the molecules of the component gases. This causes some of the incident laser light to scatter in all directions, with varying intensities depending on the particle type and size. Substantial information can then be obtained by looking at the intensity of scattered light. The principal advantages of this technique are that it is non intrusive, it does not alter the flow pattern, and it has high spatial and temporal resolution. Almost 99% of the light scattered by atmospheric particles is elastically scattered. Rayleigh and Mie scattering are the two types of elastic light scattering theories. The mechanism of elastic light scattering and the quantification of the scattering cross section and scattered intensity terms are dealt in Chapter 3. Mie theory was developed by the German physicist G. Mie (1908). It is in terms of complex series solutions and is valid for particles of all sizes. If the particle size becomes very large, then Mie theory can be simplified using geometric optics. However, for particles with a diameter much less (approximately 0.06 times or less) than the wavelength of incident light, the Mie theory reduces to a single term simplification called the Rayleigh theory. The amount of light a particular particle can scatter can be defined in terms of its scattering cross section. Thus a particle with a higher scattering cross section would scatter more light than a particle with a lower scattering cross section. For Rayleigh/Mie scattering, the scattering cross section is a dominant function of the particle size, wavelength of incident light and refractive index of the particle. The scattering cross section of hydrogen is about 1/5th that of the surrounding air molecules. Hence the intensity of light scattered by hydrogen is expected to be less than the intensity of light scattered by the surrounding air molecules. In the absence of hydrogen, there would be a steady signal from the atmosphere that would consist of the Mie signal from the aerosols and Rayleigh signal from the air molecules. In the presence of hydrogen, we should expect this signal to fall. The primary objective of this study is to detect the fall in scattered intensity in the presence of hydrogen. Because of safety issues, helium, which has a scattering cross section 0.015 times that of air is used for the experimentation. In order to match the scattering cross section of hydrogen, a mixture of 20% helium and 80% nitrogen is also used. The scattered intensity is directly proportional to the scattering cross section. The proportionality constant depends on the experimental conditions. For a fixed experimental set up, the scattered signal from a given control volume would be the same if the product of scattering cross section and number density of species in that control volume is identical. Hence, as seen in Chapter 3, the scattered signal from a mixture of 20% helium and 80% nitrogen and from pure hydrogen are the same. A second objective is to test the limits of detection. The measurements are taken at four different downstream locations for this purpose. Also measurements are taken in back scatter to test the feasibility of the technique for field measurements. Two nozzles with diameters of 14" (6.3mm) and /2" (12.6mm) are used for simulating the leak. A continuous wave low power argon ion laser is used with the 4" nozzle and a high power pulsed Nd:YAG laser was used with the 12" nozzle. For each nozzle diameter the data is recorded for the two cases of pure helium and the mixture of 20% helium and 80% nitrogen. The Reynolds number is fixed at 500 for each case and 5 Froude numbers are calculated as discussed in Chapter 3. For all combinations of Reynolds and Froude numbers the leak is an axisymmetric vertical buoyant jet. The mean and fluctuating temperature and concentration profiles of a buoyant jet are well established. Since the leak is a buoyant jet it is expected that the variation of recorded scattered voltage would follow these profiles. CHAPTER 2 LITERATURE REVIEW Mie Scattering Mie theory is the general solution for scattering of a plane electromagnetic wave by a particle of arbitrary size. In 1908 Mie first derived the relations for calculating various scattering characteristics of electromagnetic radiation, by homogenous, absorbing spheres of any diameter. The Mie solution was obtained via the solution of the wave equation which originates from the more fundamental Maxwell relations. The details of the Mie solution from these basic equations have been dealt with in references such as Kerker (1969), and Bohren and Huffman (1983). The solution involves series expansions called the angular intensity distribution functions. These distribution functions are complex and involve the calculation of Legendre polynomials and RiccatiBessel functions. Over the past 30 years numerous algorithms and subroutines have been developed mostly in FORTRAN and C to calculate these functions. The first subroutine for calculating the scattering functions was developed by Dave (1970) and incorporates the first 10 terms of the series expansion. Wiscombe (1980) developed algorithms for calculating the intensity distribution functions. These functions were plotted as a function of scattering angle and were robust for any given combination of particle size, incident wavelength and refractive index. An interactive webpage was developed by Prahl (2000) which allows the computation of the intensity distribution functions at any value of scattering angle for input values of incident wavelength, particle size and refractive index. This subroutine is used for the theoretical calculations of Mie scattered intensity in this study. The results of the subroutine are verified against published values of the angular intensity distribution functions in Kerker (1969) .The comparison is discussed in Chapter 3. McCartney (1976) lists standard aerosol distribution for continental and maritime distributions. Using these particle distributions, the number of particles of a particular size can be calculated as discussed in Chapter 5. Knowing the particle size distribution, the intensity distribution functions can be calculated using the subroutine by Prahl (2000). The details of these calculation are shown in Chapter 5. Mie scattering from particles has been used as a probe for monitoring concentration fluctuations. This technique called Marker nephelometry uses light scattered from seeded particles in a flow (Mie scatters) as a concentration probe. Since its introduction in 1961 by Rosenweig et al. (1961), this technique has proven to be a very useful probe for monitoring real time fluctuations manifested by seeded particles in the flow. Becker et al.(1967) and Shaughnessy and Morton (1977) have described the application of this technique. The same technique has been used for 2D measurements by Long et al.(1981). These workers used a plane of light to illuminate particles in a flow field and used a television camera to get a digitized 2D image of turbulent mixing. Rayleigh Scattering For particles smaller than the incident wavelength (diameter < 0.06 wavelength) only the first term of the Mie solution is needed to predict the intensity of scattered light. This single term simplification called the Rayleigh theory has been extensively discussed in literature notably by McCartney (1976) and Kerker (1969). Van de Hulst (1981) addressed the issue of assuming Rayleigh scattering to be single and independent of surrounding scatters. At 1 atm pressure and at a temperature of 300 K, air molecules are separated by distances over 600 radii. He estimated that 3 radii distance between surrounding scatters is sufficient separation to ensure independent scattering. This means that the assumption of no multiple scattering is valid for air molecules for pressures much higher than atmospheric pressures. Rayleigh light scattering is an ideal probe for gas temperature and concentration measurements since it is nonobtrusive, direct and has high spatial and temporal resolution. Using the Rayleigh scattered signal the number density of species under consideration can be calculated and since for an ideal gas at constant pressure, the number density is inversely proportional to temperature, the temperature can be known. Using this relation MullerDethelfs and Weinberg (1979) first used Rayleigh light scattering for temperature measurements in flame speed experiments. Dibble et al. (1980) used this technique to measure temperature fluctuations in premixed flames and also demonstrated that this technique could be extended to turbulent diffusion flames where the fuel and air have been carefully chosen to have identical Rayleigh scattering cross sections. Pitz et al. (1976) use RLS to measure temperature in a hydrogenair flame. Horton and Peterson (1999) carried out transient temperature measurements in an ideal gas using laser induced RLS. Flow visualizations and transient temperature measurements were done in an axisymmetric impinging jet in a rapid thermal chemical vapor deposition reactor using RLS by Matthew and Peterson (2002). Robben (1975) evaluated the spectral broadening of Rayleigh scattered light to derive a temperature in turbulent flow measurements. Rayleigh light scattering has also been used for monitoring the concentration fluctuations which occur in isothermal turbulent flows by Graham et al. (1974) and Dyer (1979). Pitts and Kashiwagi (1983) used RLS for the study of turbulent mixing. Bryner and Pitts (1992) used RLS for combustion studies. One of the major disadvantages of RLS is background glare, which is at the same wavelength as that of the scattered beam and is impossible to filter from the signal. Glare minimization is possible by blackening the surfaces. Otugen (1993) used a dual line detection RLS technique for gas temperature measurements in which they eliminated surface scattered laser light from the Rayleigh signal by using two wavelengths. A primary assumption in this study was that the ratio of the surface reflection at two wavelengths is constant. The results indicated that accurate temperature measurements were possible even when the laser light background intensity was twice the Rayleigh signal. Buoyant Jet Theory As previously stated, in this study the leak is created using pure helium and a mixture of 20% helium and 80% nitrogen. The density of both pure helium and the mixture of helium and nitrogen is different from the surrounding fluid (air). The Reynolds number is set at 500. As discussed in Chapter 3 the leak for both cases could be assumed to be an axisymmetric vertical buoyant jet. The mean and fluctuating temperature and concentration profiles of a buoyant jet are well established in numerous references notably Rodi (1982), Chen and Rodi (1980) and Schlichting (1979). In this study, the recorded scattered voltage and standard deviation profiles were compared with these well established concentration profiles. CHAPTER 3 THEORETICAL FRAMEWORK Elastic and Inelastic Light Scattering There are two types of light scattering mechanisms: elastic scattering and inelastic scattering. In inelastic scattering there is a loss in energy of the incident wave and the scattered wave is emitted at a frequency different from the incident wave i.e hVincident hvscattered. One of the types of inelastic light scattering is termed as Raman scattering and it involves a change in either the vibrational or rotational quantum number of the constituent molecule. In elastic scattering of light, there is no loss of energy between the incident and the scattered wave. i.e hVincident= hVscattered. Elastic scattering is 2 to3 orders of magnitude greater than inelastic scattering. This is the primary reason for choosing elastic scattering as a measurement technique since the signal strength is expected to be 2 to 3 orders of magnitude higher than the inelastic signal making it more easily discernible. The mechanism of elastic scattering is discussed in detail in below. Elastic Scattering Mechanism Consider an electromagnetic wave traveling through atmosphere. Scattering occurs whenever it encounters an obstacle in its path. This obstacle could be a gas molecule, dust particle or aerosols. For simplification, the term molecule is used in this description of the mechanism of elastic scattering. A molecule can be considered a mechanical oscillator carrying unequal masses and opposite charges at the center and periphery. The elastic scattering theory assumes that the molecules are non polar. This means that the negative charge is uniformly distributed over the periphery and can be assumed to be at the center. Hence the electric dipole moment, which is the product of the charge and the separation distance, is zero in its stable state. In the presence of an electromagnetic wave the charges are forced apart due to the external electric field of the wave and an induced dipole moment is created. Since the field strength of the external electric field varies periodically, the induced dipole oscillates synchronously with the field. This oscillating dipole then emits a secondary wave at the same frequency as that of the primary wave. This secondary wave is the scattered wave. Mie Theory Figure 31 shows the scattering geometry for Mie scattering. 2 Figure 31. Mie scattering geometry. The Mie theory describes the scattering of a plane electromagnetic wave by a particle of arbitrary size. The Mie theory originates from the exact solution of scattering of an electromagnetic wave equation (derived from the Maxwell relations) by a particle and has been discussed in detail by Kerker (1969). A scattered wave is generated whenever a plane wave is incident upon a particle possessing a discrete boundary and a refractive index different from the surrounding medium. In spherical coordinates the wave equation can be described as [1 Or2 8 + 1 8 sinO 8 + 1 2 + k = 0 (3.1) rr2r r r2 siniO 0 0 r2 sin2 O(p2 The solutions to this equation are the Hertz Debye potentials which can be obtained by the method of separation of variables as follows: V = R(r)O(9)D((p) (3.2) Each of these functions satisfies the following ordinary differential equations: d2rR(r) + [k2 n(n+)] rR(r) =0 (3.3) dr2 2 1 d (sinO d0(_)) + [n(n+l) m2 ] 0(0) = 0 (3.4) sinO dO dO sin20 d2Dj() + m2((P) = 0 (3.5) d(p2 where n and m are integers. The solutions of equation 3.3 are the Riccati Bessel functions defined as n(kr) = (tkr/2) 2 Jn +12 (kr) (3.6) (kr) = (tkr/2) / Nn + 1 (kr) where Jn + 12 (kr) and Nn + 12 (kr) are the half integer order Bessel and Neumann functions. The solutions of equation 3.4 are the associated Legendre polynomials given by 0 = Pn(m)(cos) (3.7) The solutions to 3.5 are the sin(myp) and cos(mp). The general solution of the scalar wave equation (3.1) (the Hertz Debye potentials) can be obtained from a linear superposition of all of the particular solutions. The Hertz Debye potentials represent the solution for the incident wave, the scattered wave and the wave inside the particle. Only the Hertz Debye potentials for the scattered wave are discussed here. The Hertz Debye potentials for the scattered wave can be expressed in terms of an infinite series and are called the angular intensity distribution functions Ii and 12. I1 and 12 are proportional to the perpendicular polarized and parallel polarized components of the light scattered at an angle 0 respectively. n=oo 11 = I Z 2n+l (anrnC(cos0)+ bnTn(cos0))2 (3.8) n=l n+1 n=oo 12 = I Z 2n+l (anTn(cos0) + bn7n(cos0))2 (3.9) n=l n+1 where 7,(cos0) = PnlcosO) (3.10) sinO and n((cos0) = d P(l)(cos0) dO The constants an and bn are obtained from the boundary conditions that the tangential components of the electric and magnetic field of the incident wave are continuous over the entire surface of the sphere. If the number density (N) of the particle is known then the Mie scattering cross section for a single particle size can be defined as oMie = )2 N(11+I2) (3.11) 872 The Mie scattering cross section is an indication of the intensity of light that would be scattered from a particle of arbitrary size. Calculation of intensity distribution functions Ii and 12 From equations (3.8) and (3.9) it can be seen that the intensity distribution functions Ii and I2 are in terms of complex infinite series and involve the calculation of the Legendre polynomials for every value of n. Also the constants an and bn involve the calculation ofRiccati Bessel functions for every value ofn. As stated in Chapter 2, subroutines for the calculation of Iand 12 are available. In this study one such program developed by the Oregon Medical Laser Center is used. The input parameters are the incident wavelength, particle size and refractive index. McCartney (1979) states that the refractive index of crystalline haze aerosols can be assumed to have a value of 1.33. The aerosols are assumed to be dielectric. This value is used throughout this study. It is to be noted that the angular intensity distribution functions depend on the refractive index and the value of 1.33 imposes a limiting condition since it does not take into account dry particles like soot. The values of I1 and I2 obtained from this subroutine are compared with published values of I and I2 in McCartney (1979). The comparison is done for five combinations of incident wavelength, refractive index and particle size. A typical comparison is shown in figures 32 and 33. Size parameter gives the relation between the size of the particle and the wavelength of incident light and is defined as a = 27a (3.12) Figure 31 shows the values of I and I2 obtained as a function of scattering angle for a size parameter (a) of 0.5 (particle radius of 0.044 microns) and refractive index of sphere of 1.33. Using the program and figure 32 is obtained from McCartney for the same input parameters. The two figures are identical, establishing the accuracy of the subroutine. 8.00E04 6.00E04 . \\ /  perpendicular 4.00E04  unpolarized 2.00E04 parallel parallel 0.OOE+00 0000000000 Nl C CO C N T COO Observation angle (deg) Figure 32. Ii (perpendicular) and I2 (parallel) intensity distribution functions for size parameter =5, and refractive index =1.33 obtained from the interactive webpage. S a 0.5 2 r = 0.044 nm 104 I l Il 0 20 40 60 80 100 120 140 160 180 Observation angle, 0 (deg) Figure 33. Ii (perpendicular) and I2 (parallel) intensity distribution functions for size parameter =5, and refractive index =1.33 from McCartney. Rayleigh Theory The Mie solution is a complex mathematical solution. For particles of size much less than the wavelength of incident light, the Mie series solution converges in one term and is called the Rayleigh theory. The Rayleigh theory was originally put forth by Lord Rayleigh (J.W. Strutt, third Baron of Rayleigh) in 1871, long before the Mie solution was developed (1908). Later on it was proved that the Rayleigh theory is actually a single term simplification of the Mie theory. Lord Rayleigh put forth the Rayleigh theory principally to explain the blue color of the sky. He assumed that the particles were spherical, isotropic, much smaller than the wavelength of incident light and denser than the surrounding medium. Through straightforward dimensional reasoning he arrived at the conclusion that scattering varies directly with the square of the particle volume and inversely as the fourth power of the wavelength of incident light. The scattering by gas molecules is in the Rayleigh regime because of the small size of the molecules. McCartney (1976) gives a good physical description of Rayleigh scattering. He states that for the Rayleigh theory to be applicable, a 1 or the particle radius should be at least 0.03 times less than the wavelength of incident light. The following are the assumptions about the molecules for Rayleigh scattering. 1. The molecules are non ionized implying that there is no overall charge over the entire molecule. This means that the molecule does not experience a net force in an electric field. 2. The molecules are non polar meaning that the electronic charge is uniformly distributed over the shell and could be treated to be at the center. Even though the polar assumption is made in the original theory, the Rayleigh theory is valid for nonpolar particles as well. 3. The molecule is isotropic implying that the forces experienced within the molecule are balanced. 4. The molecule is linear which means that the binding forces within the molecule obey Hooke's law. 5. The molecules are lightly damped meaning that the amplitude of oscillation does not become too large at frequencies near resonance. These assumptions are applicable to ordinary gas molecules like nitrogen and helium. Based on these assumptions McCartney (1976) derives the differential Rayleigh scattering cross section for perpendicular polarized scattered light as oRayleigh = 1285 a6 n2 1 (3.13) 34 n2 + 2 As seen from equation (3.13) the Rayleigh scattering cross section is independent of the scattering angle and has an inverse fourth power dependence on the wavelength of incident light. It also depends on the refractive index (n) of the particle. Photon arrival rate calculations for Rayleigh and Mie theory Knowing the Rayleigh and Mie scattering cross sections the intensity of scattered light can be calculated for a given set of experimental constants. Both the Rayleigh and Mie scattering cross sections are applicable for an individual molecule and are not functions of the gas number density N assuming independent scattering. For a given volume of a gas the intensity of scattered light is linearly proportional to the gas number density. The scattering from multiple molecules in a given volume can be considered to be additive, independent and incoherent because of the random spacing and thermal motion of the gas molecules. Thus for an incident beam of energy Io ,scattered from a control volume with molecular number density N, the intensity of the scattered beam is proportional to the scattering cross section of the molecule, the number density and the energy of incident beam Iscat = C(IooN) (3.14) The proportionality constant is defined by the scattering geometry, namely the solid angle of the collection optics (Q), control volume (dV) and optical efficiency of the collection optics (r). In this study, the scattered beam was collected using a 60 mm diameter lens at a distance of 250 mm from the scattering volume which define the solid angle. The control volume depends on the angle at which the scattered beam is collected and is discussed in chapter 4. The optical efficiency was assumed to be 90% for each optical surface the scattered beam is passed through. Thus for Rayleigh scattering the intensity of scattered light from a control volume containing a mixture of gases is given by IRayleigh (Io)(l)(dV)(Q)1(NGRayleigh)i (3.15) And scattered intensity for Mie scattering is given by Imie = (Io)(rl)(dV)(Q)X(NGMie)i (3.16) Knowing the Rayleigh and Mie scattered intensities, the total scattered intensity received at the collection lens can be calculated as Itotal = IRayleigh + IMie (3.17) Scattering Cross Section Considerations In this study, pure helium and a mixture of 20% helium and 80% nitrogen are used. The scattering cross section of helium is 0.015 times the scattering cross section of air and the scattering cross section of pure hydrogen is 0.23 times the scattering cross section of air. In order to match the scattering cross section of hydrogen, a mixture of 20% helium and 80% nitrogen is used. The differential scattering cross sections of hydrogen, helium and nitrogen are listed at a wavelength of 532 nm in table 31. Table 31. Scattering cross section at a wavelength of 532 nm. Gas Scattering Cross Section (a) (m2/sr) Air (Nitrogen) 8.16E32 Hydrogen 1.88E32 Helium 1.22E33 Knowing these values, the mixture concentration of 20% helium and 80% nitrogen are obtained using the following relation (x = % helium) Hydrogen = X Ghelium + (1X) Gnitrogen. (3.18) Buoyant Jet Theory Buoyancy forces arise in a jet if the density of the flow fluid is different from the density of the surrounding fluid. In the absence of buoyancy forces the jet is called a non buoyant jet. In the other limiting case when the buoyancy force dominates the flow, the jet is called a plume. Thus the non buoyant jet has about the same density as the surrounding environment so that the buoyancy forces are absent, whereas a pure plume has no initial momentum. The densities of both flow fluids used for the experiments pure helium and the mixture of helium and nitrogen are different from the density of the surrounding fluid (ambient air). The Froude number is used to characterize whether a jet is a non buoyant jet, a buoyant jet or a plumeChen and Rodi (1980). The Froude number is the ratio of inertial forces to buoyancy forces and is defined as Fr = Re/Gr2 (3.19) Reynolds number is the ratio of the inertial forces to the viscous forces and can be written as Re = vd (3.20) The Grashof number (G) is defined as g(f2ao)D3 (3.21) po) And it is the ratio of buoyant to viscous forces. In a non buoyant jet, only the Reynolds number is of influence (Fr=oo) whereas in pure plumes only the Grashof number is dominant (Fr=0). For an axisymmetric vertical jet, the limiting condition for a buoyant jet is defined in Chen and Rodi as follows 0.5 < Fr1/2 (po/pa1/4 (z/D) < 5 (3.22) Froude Number Calculation Two different nozzle diameters of 1/4" and 1/2" are used. Also two flow fluids (pure helium and mixture of 20% helium and 80% nitrogen) are used. Four different cases are considered: 12" diameter nozzle with pure helium, 12" diameter nozzle with a mixture of 20% helium and 80% nitrogen, 14" diameter nozzle with pure helium and 14" diameter nozzle for a mixture of 20% helium and 80% nitrogen. The measurements are taken for four downstream distances of 2, 4, 6, and 8 nozzle diameters. The Reynolds number is chosen as 500 for each case. The criterion for buoyant jet (equation 3.21) is tested for all four combinations of nozzle diameters and flow fluid and at all four downstream distances. Table 32 lists the values of Froude numbers for all four combinations of nozzle diameters and flow fluid and for the two limiting cases of 2 and 8 nozzle diameters downstream. Table 32. Froude number calculations show that the criteria for buoyant jet is met for all combinations of flow fluids, nozzle diameters and downstream distances Cases Reynolds Froude 0.5 < Fr1/2(o/a)1/4z/D <5 Number(Re) Number(Fr) 2 nozzle 8 nozzle 2 nozzle 8 nozzle diameters diameters 1. 1/2 "; He 500 36.5 0.521 2.08 2. 1/2"; 20%He, 500 64.5 0.54 2.16 80%N2 3. /4"; He 500 292.5 0.88 3.52 4. /4 "; 20%He, 500 516.5 0.94 3.76 80%N2 Buoyant Jet Profiles and Concentration Variation Figure 34 shows the instantaneous and time averaged profiles of a buoyant jet. The jet centerline is characterized by a potential core near the nozzle exit. Inside the potential core, the concentration of the flow fluid is 100%. The shear layers define the jet spread angle (P).The edge of the shear layers mark the boundary of the time averaged profile. The concentration of the flow fluid varies from a 100% at the jet centerline to 0 % at the edge of the shear layers. Chen and Rodi (1980) have listed the spread angles of non isothermal jets. In this study a spread angle of 130 is found for a vertical round buoyant jet. Chen and Rodi (1980) summarize empirical data predicting the axial spread of an axisymmetric vertical buoyant jet. The concentration at any point in the shear layer of a buoyant jet can be calculated using the following relation. C Ca = exp [Ko(r/z)2 ] (3.23) Co Ca tme Averaged Profile Shear layer Instantaneous profile Potential core Black cover Figure 34. Instantaneous and time averaged profiles of a typical buoyant jet. The constant Kc is related to the jet spread angle 0 as Kc = In 2/(tanp/2)2 So, for ajet spread angle of 130 Kc = 53.4 (3.24) For a vertical buoyant jet, the centerline concentration (Cc ) is found from the dimensionless density (C ) using the following equation Co Ca= C* = 4.4Fr l/(po/pa) 7/16 (z/D)5/4 Cci Ca (3.25) Using this value of Kc the concentration of the flow fluid at any point in the flow can be calculated using equation 3.22. For both the 1/2" diameter nozzle and the /4" diameter nozzle, C Ca/Ccl Ca is plotted as a function of radial distance (r) for a downstream distance (z) of 2 nozzle diameters for each case as seen in figure 35. Figure 35. Concentration variation with r for z = 2 is Gaussian in nature for both nozzle diameters of 14 and /2". 1.2 1 0.8 r 0.6 0.4 0.2 0 1/4 inch nozzle 1/2 inch nozzle So r qt . Wo LON C 0 o O I% I W o S. ri Ci i 4 4 ..6 i 6 r r CHAPTER 4 EXPERIMENTAL SCHEME Collection at 90 and Backscatter The goal of the project is to detect the change in intensity of scattered light in the presence of hydrogen. Rayleigh scattering is independent of the angle at which the scattered light is collected but the Mie signal depends on the scattering angle. Initially, the collection optics were set at 900 to the incident beam and the measurements were focused on a single point of the beam as seen in figure 41. An advantage of this optical configuration is that the glare from the incident beam on the scattered signal is minimum at this angle since the minimum incident beam area is visible to the photomultiplier tube. For field measurements since the exact position of the leak could be at any distance from the laser, the drawback with a 900 scattering scheme would be that the collection optics consisting of the collecting, focusing lenses, filter, photomultiplier tube and the digital oscilloscope would have to be moved depending on the position of the suspected leak. It would be ideal to use a scheme in which the collection optics could be kept stationary. This can be done by using a backscatter scheme in which the incident light is passed between two 4" (101.6mm) square mirrors and the scattered light is collected at a 1800 using the two mirrors. The mirrors are 90% reflective at a wavelength of 532 nm. The backscatter scheme is shown in figure 42. For the backscatter scheme, since the angle of observation is a 1800, the control volume is defined by the beam area and the pulse width of 5 ns (1.5m). This leads to a fall in the signal to noise ratio as the leak occupies only 0.6% of the control volume as discussed in Chapter 5. Beam dump  40 4 Collecting lens 4 Focusing lens I Band pass filter 4 Photomultiplier tube Figure 41. Experimental scheme for collection of scattered light at 900 The back scattering scheme uses the same set of collection optics as that for the 900 scattering scheme except that the 900 scheme uses a 60 mm collecting lens. From ray tracing it is found that the beam divergence angle of the scattered beam arriving at the Laser Nozzle mirrors was 0.40 and so the scattered rays reflected from the mirrors are effectively collimated and could be focused on the photomultiplier tube. I4 F Beam dump Nozzle Focusing lens Band pass filter 4 Photomultiplier tube Figure 42. Experimental scheme for collection of scattered light at a 1800 (back scatter) The following components constitute the optical set up shown in figures 4.1 and Laser Mirrors Laser. Initially the experimentation was done using an argon ion laser operated on the 488 nm wavelength. The laser power was set at 3 W and the beam diameter was 1mm with a beam divergence angle of 0.060. The argon ion laser is a continuous beam low power laser. In order to amplify the scattered signal and for higher temporal resolution a pulsed Nd:YAG laser was also used. The laser power is 200 mJ/pulse at a wavelength of 532 nm. The pulse width was 5 nanoseconds and the frequency is 10 Hz. The beam diameter is 6 mm with a beam divergence angle of 1.1. Since the beam diameter was larger than the ArgonIon laser, the Nd:YAG laser defined a larger control volume than the ArgonIon laser. Thus two different power lasers at two wavelengths are used for the experiments. Nozzle. The leak was simulated using a nozzle. The laser beam was passed directly over the nozzle. The nozzle is placed at a distance of 8 feet (20.32 cm) from the laser for the 900 scattering scheme. Using the beam divergence angle, the beam diameter over the nozzle was calculated to be 1.1 mm for the Argon ion laser and 8 mm for the pulsed Nd:YAG laser. In order to analyze the effect of the relative size of the beam and leak, two different nozzle diameters are used. For initial measurements using the argon ion laser a 1/2" (12.6mm) nozzle was used. For the pulsed laser a /4 (6.3mm) diameter nozzle was used. Thus for the argon ion laser, the beam diameter is 11.5 times less than the leak diameter whereas for the Nd:YAG laser the beam diameter is 1.25 times more than the leak diameter. The back scattering scheme uses the pulsed Nd:YAG laser with the 14 diameter nozzle. The nozzle is placed at a distance of 20 feet (50.8 cm) from the laser and the beam diameter is 10 mm over the nozzle. The Reynolds number of the leak is set at 500 for both nozzle diameters and the Froude number was calculated as discussed in chapter 3 and presented in table 3.1. This combination of Reynolds and Froude number ensured that the leak is a buoyant jet. The nozzle is covered with a black felt drape to minimize glare. The mass flow rate of pure helium and the mixture of helium and nitrogen is monitored using mass flow controllers. Nozzle mounting. Three micrometer traverses are used for mounting the nozzle for three dimensional motion. Each traverse has a least count of 0.05 mm. Figure 43 shows the schematic of the mounting used. This allowed the measurements to be taken at different downstream locations. r Nozzle y z Traverse 3 Y Traverse 2 Traverse 1 Figure 43. Nozzle mounting showing three dimensional motion capability. Beam dump. The incident laser beam was trapped using a beam dump. The beam dump was placed at a distance of 20 feet from the nozzle to minimize glare. Collection optics. The scattered beam was collected using a set of collection optics which consisted of a collecting lens, focusing lens, band pass filter and a photomultiplier tube. The collection optics were covered with a black felt drape in order to eliminate additional glare. Collecting lens. The collecting lens is a 60 mm diameter lens and has a focal length of 250 mm. As seen from figure 41 the focal point of the lens coincides with the nozzle position so that scattered light collected by the lens is collimated. As seen from figure 42 the collection lens was not used for the backscattering geometry because the scattered beam reflected from the two mirrors was collimated. Focusing lens. The focusing lens is a 60 mm diameter lens and had a focal length of 124 mm. The collimated beam from the collecting lens is focused on the Photomultiplier tube using the focusing lens. Band pass filter. For the Nd:YAG laser, the band pass filter is a 532 nm line filter that blocks background light at other wavelengths from reaching the photomultiplier tube. The band pass filter was kept at a distance of 3 mm from the Photomultiplier tube. Photomultiplier tube. The scattered beam is focused on the Photomultiplier tube. The Photomultiplier tube is a Hamammatsu model number HC 12001 tube and has a built in amplifier with adjustable gain. It has a rise time of 2 ns and a bandwidth of 23 Khz. This means that the photomultiplier tube can distinguish between two signals that are 44 microseconds apart. Since the frequency of the pulse laser is 10 Hz the interval between two consecutive pulses is almost Ims. Hence for a 900 scattering geometry, the photomultiplier tube can distinguish between two consecutive pulses. From the manufacturer's specifications the calibration constant of the photomultiplier is found to be 121 V/nW which allows the conversion of the voltage recorded on the photomultiplier tube into power. The Photomultiplier tube has a 0.015 mm optical slit mounted on it to minimize the intensity of scattered light. The Photomultiplier tube converts the photonic signal to electrical voltage which is then sent to a high speed digital oscilloscope. It is important to note that even though the photomultipliers are considered to be highly linear devices, the linearity generally occurs over a lesser range for the pulsed Nd:YAG laser. (X = 532 nm) at varying flash lamp discharge voltage. Figure 44 displays the photomultiplier tube voltage as a function of varying flash lamp discharge voltage. 680 660 660 R2 =0.9931 640 620 600 * Series1 580 Linear (Series1) 560 540 520 500 0.9 1 1.1 1.2 1.3 1.4 Figure 43. Photomultiplier tube linearity tested as a function of flash lamp voltage High Speed Digital Oscilloscope. The photomultiplier tube signal is recorded on a high speed digital oscilloscope (LeCroy). It has two channel simultaneous data acquisition capabilities. It is triggered externally using the pulsed laser. The trigger time is 180 ns which means that the laser sends an electric signal to the oscilloscope exactly 180 ns before it fires the pulse so that the oscilloscope can be set to capture the scattered signal. The oscilloscope acquires data over a time period of 50 microseconds. The measurements are recorded and readout on a spreadsheet. Control Volume. For the 900 scheme, the control volume is defined by the beam diameter and the length of the control volume is defined by the width of the optical slit (150 microns). Back Scatter Considerations As previously stated the back scattering scheme is used to test the feasibility of the technique for field measurements. An important consideration for field measurements is the ability to detect the leak over longer distances. Hence the leak is created by placing the nozzle at a distance of 20 feet (50.8 cm) the laser beam. The beam dump is placed at a distance of 10 feet (25.4 cm) from the nozzle. The control volume is defined by the beam diameter and the length of the control volume for this case is defined by the pulse width (5nanoseconds =1.5m) Data Recording Procedure The measurements were done at four different downstream locations of 2, 4, 6 and 8 nozzle diameters for the pulsed Nd:YAG laser and two downstream locations of 2 and 6 nozzle diameters for the argon ion laser. For each downstream distance the nozzle was traversed along the direction of the beam using the micrometer traverse as shown in figure 43. The steps in which the nozzle was traversed was calculated from the spread angle of the jet as seen in figure 44. The distances la, 2b, 3c and 4d varied depending on the nozzle diameters (1/4" or 1/2"). This was done so that the measurements done at all four downstream distances could be graphed on the same scale of r/z. d 4 c 3 b 2 8d r 1 6d y 4d a 14 S2d 6.50 5 nozzle Figure 44. Calculation of nozzle traverse distances for downstream distances of 2, 4, 6 and 8 nozzle diameters using jet spread angle of 130. CHAPTER 5 RESULTS AND DISCUSSION Comparison of Theoretical and Experimental Photon Arrival Rate This section discusses how the theoretical and experimental photon arrival rates compare. In chapter 3 the equations for calculating the Rayleigh photon arrival rate (equation 3.15) and Mie photon arrival rate (equation 3.16) are presented. The total theoretical photon arrival rate is calculated as a sum of the Rayleigh and Mie photon arrival rates (equation 3.17). Calculation of Theoretical Rayleigh Photon Arrival Rate. In the presence of helium, the Rayleigh photon arrival rate can be calculated by modifying equation 3.15 as follows. The percent helium (x) was varied from 0 to 100 and the resulting photon arrival rate is shown in figure 5.1. IRayleigh = (Io)(rl)(dV)(Q)N(xHe + (1x) Gair) (5.1) Io = 7.07E+6mj/m2pulse = 1.88E22 photons/ m2pulse(Nd:YAG laser); 3.8E+6 W/ m2 (ArgonIon laser) f = (0.9)5 dV = 1.2 E8m3 (900 scatter); 1.2E4m3 (backscatter) Q = t(60)2/(250)2 sr (900 scatter); (2)(101.6)2/(12100)2 sr (backscatter) N = 2.2E25 molecules/m3 OHe= 1.22E33 m2/sr Gair = 8.16E 32 m2/sr _ 8.20E+09 7.20E+09 S6.20E+09 Rayleigh photon 0 5.20E+09 arrival rate S9 Mie photon arrival S4.20E+09 IM rate S3.20E+09 total photon arrival rate 'I 2.20E+09 rate  1.20E+09 =. 2.00E+08 0 10 20 30 40 50 60 70 80 90 100 % Helium Figure 51. Theoretical photon arrival rate calculations show that Rayleigh signal is higher than Mie signal. Calculation of Theoretical Mie Photon Arrival Rate Equation 3.16 is used for the calculation of the Mie photon arrival rate. IMie = Io)(l)(dV)(Q)1(N, Mie)i (3.16) The two unknowns in this equation are Ni and (oMie)i The calculation of (oMie)i involves the use of the intensity distribution functions II and 12 If the particle size is known, these functions can be calculated for a give wavelength of incident light using the subroutine as discussed in Chapter 3. Two size distributions were used to determine the size of particles in the ambient air and also the number density of particles of a particular size. Distribution 1 uses the data from a LASAIR II particle counter. Table 51 shows a typical output of the particle counter data. The sampling is done in the laboratory for a sampling time of 1 minute and the volume of particles sampled over this period is 1 cubic feet. The particle size is given in column 1. Columns 2 and 3 give the lower and upper limits of the number of particles of the corresponding size. An average of columns 2 and 3 is taken. Thus using the LASAIR II data, the number densityparticles per m3 (N i) of a particular size are known. For each particle of size of 0.3, 0.5, 1, 5, 10, 25 microns the Mie scattering subroutines are used to calculate II and 12 at a wavelength of 532 nm for the pulsed Nd:YAG laser. Knowing II and 12 ,oMie is calculated using equation 3.11 and the Mie scattered intensity using equation 3.16. To account for the presence of helium equation 3.16 was modified as lMie= IoTllQNGMie(1xHe) (5.2) Table 51. Particle counter data show the particle distribution in the lab. Number of particles/ft3 Number of particles/ft3 Particle size (microns) (lower limit) (upper limit) (lower limit) (upper limit) 0.3 25268 27272 0.5 7864 9222 1 1164 1298 5 100 104 10 12 12 25 1 1 For distribution 2 the typical maritime aerosol distribution in McCartney (1969) is used. Figure 52 shows particle distribution for stratospheric dust particles or hailstones (model H), continental aerosols (model L) and maritime aerosols (model M). The maritime aerosol distribution (model M) is chosen since it is appropriate for Cape Canaveral. The number density per radius interval n(r) for this distribution is calculated using the following fit (McCartney) n(r) = ar" exp(br7) (5.3) Where the constants a, b ,a, y have the following empirically determined values a = 5333 b = 8.9443 a=l y =1/2. io3   102 2 H L M  102 (3. 1071 1 0.01 0.1 1.0 Particle radius (pm) Figure 52. Model M shows maritime distribution of aerosols which is used to calculate the particle size distribution.(McCartney 1979, page 139) Using the values of particle size (r) ranging from 0 to 25 microns, and the above values of the constants, the model M (loglog scale) was duplicated on a regular scale as shown in figure 53. 350 300 o 250 200  S200 Model M S150 o 2 100 M 50  0 0.1 1 10 100 particle size Figure 53. Model M duplicated for calculating the number density using radius interval on a lognormal scale. The number density is calculated for the same particle radii as the LASAIR II data. (i = 0.3, 0.5, 1, 5, 10 and 25 microns) by integrating the area under the curve shown in figure 52 using following equation. The interval (i) to (i+1) represents the difference between the two consecutive particle radii i=6 Ni = 0.5(n(r)i + n(r)i+1)I (5.4) i=l The values of number density Ni (particles/m3) obtained from equation are compared with the values of Ni from the particle counter (Table 5.1). Figure 54 shows this comparison. As seen from the figure the particle distribution for maritime and lab aerosols is similar. For distribution 2 since the number density and the particle radius are known (equation 5.4) the Mie scattering cross section and the intensity of Mie scattered signal is calculated using the same approach as in distribution 1. 800 700 600 = 500 S400 E 300 Z 200 100 0 0.1 McCartney * Lab 1 10 particle size Figure. 54. Comparison of number density of maritime and lab aerosols shows similarity Figure 55 shows the comparison between the Mie photon arrival rates for the two distributions. The value of IO used is that for the pulsed Nd:YAG laser. 2.10E+09 1.60E+09 1.10E+09 6.00E+08 1.00E+08  Maritime distribution  lab data 20 40 60 % Helium 80 100 Figure 55. Effect of maritime and lab aerosol distribution on Mie scattered intensity shows that both signals are of the same order of magnitude. As seen from the graph, both signals are of the same order of magnitude. The maximum difference between the two signals (55%) is for the case of 0% helium since all " CL o0 I Mie scattering particles have been displaced, and no ambient air is in the control volume. Both signals tend to zero as the helium concentration approaches 100%. This figure shows that the Mie signal would be of the same order and smaller for maritime field measurements as that for laboratory measurements. Calculation of Total Theoretical Photon Arrival Rate For the calculation of IMie the data from the particle counter is used since it represents aerosol distribution for the experimental conditions. Knowing IRayleigh and IMie, the total theoretical photon arrival rate is calculated using equation 3.17. Figure 51 shows the variation of Rayleigh, Mie and total photon arrival rate with percent helium. The value of IO used here is for the pulsed Nd:YAG laser. An important result of the theoretical study is that the Mie scattered signal is lesser than the Rayleigh scattered signal. Initially it was reasoned that the Mie scatters would augment the scattered signal since in the presence of helium the total signal would fall due to reduction in Mie signal. This study shows that although this factor is present it influences the total scattered signal to a lesser degree. Calculation of Experimental Photon Arrival Rate The voltage as a function of time waveform obtained on the oscilloscope represents the total scattered intensity recorded at the photomultiplier tube per pulse. Figure 56 shows a typical waveform of a burst seen on the oscilloscope. The waveform is recorded at a downstream distance of 4 nozzle diameters with no flow fluid. Integrating this waveform using equation 5.5 gives the total scattered intensity per pulse (VPMT) i=100 VPMT = 0.5(Vi +Vi+l)(0.1) (5.5) i=0 where (i) represents the time scale from 0 to 100 microseconds varied in steps of 0.1 each. 0.7 0.6 0.5 I 0.4 0.3  waveform S 0.2 0.1  0 10 20 30 40 50 60 70 80 90 0.1 time microsecondss) Figure 56. Typical waveform of a burst seen on the oscilloscope (Downstream distance of 4 nozzle diameters, no flow fluid) The Photomultiplier tube calibration constant (CPMT) at the wavelength of 532 nm is found to be 121 V/nW from the manufacturers specifications. The measurements are done using the pulsed ND:YAG laser at 6 different helium concentrations of 0, 20, 40, 60 80 and 100% using the /4" diameter nozzle. As discussed in buoyant jet theory (Figure 35) the flow fluid (helium) concentration can be predicted in the shear layer for a combination of downstream distances and radial distance (r/z). The downstream distance (z) is fixed at 2 nozzle diameters. From equation 3.23 it is seen that for a fixed value of z, the radial distance (r) can be calculated if the concentration C is known since the centerline concentration of the flow fluid (Ccl) is known using equation 3.25 and the concentration of flow fluid (Ca) in the ambient is 0%. C Ca = exp [Ko(r/z)2 ] (3.23) Ccl Ca Thus for 6 different values of C (0, 20, 40, 60, 80 and 100), the measurements are made for the corresponding values of radial distance(r) obtained from equation 3.23. Figure 57 shows the values of radial distance r where the measurements are made for each value of C. Concentration profile Point of measurement z=2 Shear layer Figure 57. Figure shows points in shear layer where measurements are made corresponding to each value of C using pulsed Nd:YAG laser and /4" nozzle at a downstream distance of 2 nozzle diameters. Re = 500; Fr = 290. For each measurement the integrated area under the voltage time curve on the oscilloscope represents the total scattered signal per burst. Equation 5.6 is used to convert the voltage recorded for each measurement into scattered signal in photons per pulse. experimental= VtPMTX/CPMThc (5.6) Figure 58 shows the comparison of the experimental and theoretical photon arrival rates for varying percentages of helium. 41 @ 9.00E+09 8.00E+09 S8.E+09 R2 = 0.9978 a 7.00E+09  0 0 theoretical photon 6.00E+09 arrival rate S5E9 experimental photon 5.00E+09  5 Arrival rate F 4.00E+09 Linear (experimental 'E 3.E+09 photon arrival rate) 2.00E+09 =C 1.00E+09 1 .0 0 E + 0 9 . . .. 0 10 20 30 40 50 60 70 80 90100 % helium Figure 58. Comparison of experimental and theoretical photon arrival rates. The experimental and theoretical photon arrival rates are of the same order of magnitude with a maximum error of 57.3% for the case of 100% helium and a minimum error of 11.3% for the case of 0% helium. A linear regression of the experimental photon arrival rate gives a coefficient of 0.99 indicating a constant additional factors) contributing to the error independent of percent helium. This constant error can be attributed to 1.Background glare, which is at the same wavelength as that of scattered signal. 2. Deviation of optical efficiency of each collection surface from the assumed value of 90%. 3. Ambient air in control volume at jet edges. Data Analysis Techniques As previously stated for each burst the scattered signal from the photomultiplier tube is recorded as a voltagetime curve on the oscilloscope. Two techniques of analyzing the recorded voltage are considered. The Nd:YAG pulse laser with the 14" nozzle is used for both techniques and the flow fluid is pure helium for both cases. Integrated area method Initially a set of 1000 data points per burst are captured from the oscilloscope and the integrated area under the voltage as a function of time curve is computed. For the measurements, the downstream distance is 2 nozzle diameters, Reynolds number is 500 and Froude number is 290. The measurements are done in the shear layer of the leak at a radial position corresponding to 60% helium concentration as shown in figure 57. This is because the maximum variation of recorded voltage is expected to be in the shear layer edge since the intermittency of the turbulence is maximum. This procedure was repeated for a set of pulses at the same nozzle position shown in figure 57 until the areas converged. Figure 59 shows the convergence studies for the area. 3  1 2 > normalizd running S0 average of area S1 6 7 8 normalized inidvidual area > 2 3 4 pulse number Figure 59. Convergence studies of normalized area and averaged area shows that the area converged in 10 pulses. Series 1 represents the running average of the integrated area normalized with respect to the mean of 10 pulses. As seen from the graph, for a measurement in the shear layer a set of 10 pulses are enough for convergence. The maximum percent variation of 43 any individual area from the mean is 3% and the maximum percent variation of the running average of area from the mean is 2.6%. Peak Voltage Method Two studies are undertaken to validate the use of peak voltage rather than integrated area as a repeatable and reproducible data measurement technique. A convergence study is also done to determine the number of peaks required for data measurement. These studies are carried out using the pulsed Nd:YAG laser with the /4" nozzle. Study 1. For four downstream positions (z) of 2, 4, 6, and 8 nozzle diameters the waveform is recorded along the jet centerline (r=0) as seen in Figure 510. 0.8 0.7 0.6 .t : 0.5 __ z/d=2 > 0.4 i z/d=4 0  0.3 i \ z/d=6 > 0.2 z/d=8 0.1  0.1 10 20 30 40 50 60 70 80 90 time microsecondss) Figure 510. Waveform with varying glare at four downstream locations. All four measurements are done in room air with no helium flowing through the nozzle. In this case we expect that the Rayleigh and Mie scattering signals are constant and the only variable in each case was the glare from the nozzle. 44 All four waveforms were normalized with the corresponding peak as shown in figure 5.11. It is seen that all four waveforms fall on top of each other and it is impossible to distinguish between them. A four way cross correlation analysis of the waveforms was done using the following relation rxy= Xi x)(yiy) (5.7) 4 ((x x)Nd)2) yY)Nd)2) The cross correlation coefficient was 99.7%. The high cross correlation coefficient supports the hypothesis that the peak voltage contains sufficient information of each waveform. 1.2 1  0.8  z/d=2 m 0.6 z/d=4 > 0.4 z/d=6 S, z/d=8 0.2  0 0 ........ ........ .. ..... .... .. 0.2 10 20 30 40 50 60 70 80 90 100 time microsecondss) Figure 511. All four waveforms normalized with their individual peaks to remove glare, the waveforms are indistinguishable. Study 2. In this study, a correlation analysis of the peak voltage and the corresponding integrated area is done for 10 pulses at the same nozzle position in the shear layer as that used with the integrated area studies above (Figure 5.7). The flow fluid is 100% helium and the Reynolds and Froude numbers are 500 and 290 respectively and radial position corresponding to 60% helium concentration. The downstream distance is 2 nozzle diameters. The correlation coefficient is 99.3% which again supports the relation between peak and area for a given pulse. Peak Convergence Study. From the previous two studies it is established that the peak contains all necessary information about the waveform and the peak and area are related. The convergence study is undertaken to establish the number of pulses required for convergence. The downstream distance is 2 nozzle diameters and the Reynolds and Froude numbers are 500 and 290 respectively for 100% helium. The measurements are done in the same position in the shear layer of the leak as for the previous two cases of integrated area and comparison of integrated area and peak (Figure 5.7) (radial position corresponding to 60% helium concentration. The average peak voltage is recorded after the 1st, 100th, 200th, 300th, 400th and 500th pulse. Table 5.2 lists the voltage recorded after each measurement. Column 3 of the table gives percent variation between two consecutive values of recorded voltage. Table 52. Shows that average peak voltage recorded after 300 pulses in shear layer is a converged mean. Pulse Number Average peak voltage (Vi) (Vi Vi)/V Pulse Number (Vi 1) (mV) 0 550 100 576 4.7% 200 564 2.1% 300 559 0.88% 400 563 0.76% 500 558 0.89% As seen from table 5.1 the variation between average peak voltage is less than 1% if the peak voltage was recorded after 300 pulses. Also these measurements are carried out in the edge of the shear layer of the jet where the fluctuations in the flow are maximum. Hence it is concluded that for any nozzle position, recording the average peak voltage after 300 pulses is a reliable data measurement technique. Analysis of Recorded Data Figure 56 shows the comparison between the predicted (theoretical) and the measured (experimental) photon arrival rates. As stated earlier these studies are carried out using the pulsed Nd:YAG laser with a /4" diameter nozzle at a downstream distance of 2 nozzle diameters. This section discusses the variation of peak voltage and standard deviation of the recorded voltage for the following four cases: * Argonion laser with a 1/2" diameter nozzle at downstream distances of 2 and 4 nozzle diameters for pure helium for 900 scattering scheme. * Pulsed Nd:YAG laser with a /4" diameter nozzle at downstream distances of 2, 4, 6 and 8 nozzle diameters for pure helium for 900 scattering scheme. * Pulsed Nd:YAG laser with a /4" diameter nozzle at downstream distances of 2and 4 nozzle diameters for a mixture of 20% helium and 80% nitrogen for 900 scattering scheme. * Pulsed Nd:YAG laser with a /4" diameter nozzle at downstream distances of 2, 6, 10 and 16 nozzle diameters for pure helium in back scatter. The nozzle was traversed in the radial (r) direction as discussed in chapter 4 for the 900 scattering scheme and in the (y) direction for the back scatter scheme. Case I: Argonion laser Peak voltage variation Initially the measurements were carried out using the argonion laser (X =488 nm) at two downstream distances of 2 and 6 nozzle diameters. The nozzle diameter of 12" is used with the argonion laser and the flow fluid is pure helium. The Reynolds number of the leak is 500 and the Froude number is 35. the nozzle traverse distance for each measurement at each downstream position were calculated as discussed in Chapter 3. Three sets of measurements were done at each downstream position. Figure 511 shows the average of the normalized peak for all 3 measurements. Also shown are the error bars at both downstream locations corresponding to the standard deviation of each measurement point. The x axis is the nozzle traverse distance (r) in mm. As seen from the figure, the scattered voltage reaches a minimum value of zero at the jet centerline for both downstream distances. Also the voltage variation inside the shear layer is visible. Hence due to the presence of helium in the shear layer, there is a fall in voltage inside the shear layer. As the downstream distance increases, the jet spreads out and the fall occurs over a wider radius. *"$ 1 11E K x]E *F E z/d=2 z/d=6 0 4 15 10 5 0 5 10 15 r (mm) Figure 512. Voltage variation for ArgonIon laser shows that there is a fall in voltage in presence of helium; 1/2" nozzle at two downstream distances; Re =500; Fr =35. As seen from figure 512, the fall in peak voltage at the jet centerline which corresponds to 100% helium. Also, the fall in voltage occurs only in the presence of helium. At the ambient where there is no effect of flow fluid on the recorded voltage, the voltage remains constant. The fall in voltage in presence of helium occurs because 1. Helium molecules have a scattering cross section which is 0.015 times the cross section of the surrounding air molecules. 2. Molecules of the flow fluid displace some aerosols in their path which are Mie scatters. Hence there is a fall in the intensity of scattered light Normalized peak voltage variation. The raw peak voltage was normalized with the centerline voltage (Vi Va)/(Vcl  Va) as shown in figure 513. The normalized concentration profile plotted in chapter 3 is also shown as a function of r/z From figure 513 it is seen that the voltage variation follows a similar Gaussian distribution as that of well established normalized concentration profile of a axisymmetric vertical buoyant jet. z/d=2 z/d=6  normalized S / concentration 2.0 1.0 0.0 1.0 2.0 r/z Figure 513. Normalized peak voltage variation for argonion laser for pure helium; Re= 500; Fr=3.5; nozzle diameter = 1/2 Standard deviation profiles Figure 514 shows the variation of the normalized standard deviation ratioed to the centerline voltage o/(Vmax Vcl) for the downstream distance of 2 nozzle diameters The downstream distance of 2 nozzle diameter represents the near field regime of the buoyant jet. It is seen that inside the potential core, the standard deviation falls as expected. 25  20  15 * S 10  5 0o 1.40 1.12 0.84 0.56 0.28 0.00 0.28 0.56 0.84 r/z Figure 514. Percent standard deviation variation for near field case showing the reduced fluctuation in the potential core. Figure 515 shows the variation of the normalized standard deviation of the data radioed to the centerline standard deviation for the far field case after the shear layers have coalesced. 50  45  40  35 30 zd=6 25 20 S15 *** 10 5 0 1.40 1.12 0.84 0.56 0.28 0.00 0.28 0.56 0.84 r/z Figure 515. Percent standard deviation for far field case after shear layers have coalesced. ed. The percent standard deviation is above the typical maximum variation of 30%. A possible reason for this is there is an additional factor of Mie scattering contributing to the increased standard deviation of the recorded voltage. Case II: Pulsed Nd:YAG laser; pure helium Peak voltage variation. Figure 516 shows the variation of raw peak voltage with respect to the nozzle position for pure helium for the downstream distances of 2, 4, 6 and 8 nozzle diameters for the pulsed Nd:YAG laser (X=532 nm) using the 14 nozzle. The Reynolds number is 500 and Froude number is 290. As seen from the graph, for each individual downstream distance the voltage tends to remain constant outside the shear layer (ambient) due to absence of helium. As observed with the argonion laser, the fall in voltage occurs over a wider radius as the jet spreads out. Also the minimum voltage is recorded at the jet centerline. i z/d=2 0 z/d=4 " z/d=6 z/d=8 0.7 30 20 10 0 10 20 30 40 r (mm) Figure 516. Voltage variation for pulsed Nd:YAG laser shows a similar profile of fall in voltage in presence of helium; Re= 500; Fr=290; nozzle diameter = 14 " Normalized Peak Voltage Variation Figure 517 shows the variation of the normalized peak voltage for all four downstream distances for the pulsed Nd:YAG laser for the case of pure helium. Also shown is the theoretical normalized concentration profile plotted in chapter 3 as a function of r/z. \ z/d=2 Sz/d=6 z/d=6 z/d=8 Normalized concentration 2 1 0 1 2 r/z Figure 517. Normalized peak voltage variation for Nd:YAG laser for pure helium; Re= 500; Fr=290; nozzle diameter = /4 " Standard deviation profiles Figure 518 shows the standard deviation profiles for all four downstream distances of 2, 4, 6 and 8 nozzle diameters. Comparing figure 518 with the standard deviation profiles for the Argonion laser (figures 514 and 515), it is seen that the standard deviation variation is less distinct for the Nd:YAG laser. A possible reason for this could be the effect of the relative size of beam and nozzle (leak) diameters. The beam diameter for the argon ion laser is 1/10th the diameter of the nozzle whereas for the pulsed laser it was 1.2 times the nozzle diameter (Figure 519). This factor was an additional variance for the pulsed laser. 20 18 x XXx 16 X X 14 XX x *z/d=2 12 *zld=4X g L M 8 gz/d=6 8~d S 6 X z/d=8 4 2 0 1.47 1.10 0.74 0.37 0.00 0.37 0.74 1.10 1.47 r/z Figure 518. Standard deviation variation for Nd:YAG laser for pure helium; Re= 500; Fr=290; nozzle diameter = 1/4 " Laser beam 1.2mm Nozzle  12.6mm Nd:YAG laser 8mm beam Laser beam Nozzle 6.3mm Argonion laser Figure 519. Relative size of beam and leak diameter for the Argonion and pulsed Nd:YAG laser. Case III: Pulsed Nd:YAG laser; 20%helium ,80% nitrogen Peak Voltage Variation Figure 520 shows the variation of peak voltage for a mixture of 20% helium and 80% nitrogen for a downstream distance of 2 and 4 nozzle diameters for the pulsed Nd:YAG laser. This mixture of helium and nitrogen matches the scattering cross section of hydrogen as discussed in Chapter 3 and hence the amount of light scattered by the mixture is expected to be identical to that of Hydrogen. The Reynolds and Froude number were 500 and 515 respectively. ii z/d=2 m z/d=4 0.92 . 15 10 5 0 5 10 15 20 Figure 520. Voltage variation for the pulsed Nd:YAG laser for the mixture of helium and nitrogen shows that there is a fall in voltage; nozzle diameter 1" ;Re=500, Fr = 515. The fall in voltage from figure 520 is 30 mV (6%) for the downstream distance of 2 nozzle diameters. The fall in voltage for z/d=2 for the case of 100% helium is 155 mV (30 %) as seen in figure 516. Normalized peak voltage variation. Figure 521 shows the normalized peak variation for the mixture at the downstream distances of 2 and 4 nozzle diameters. A comparison with the normalized concentration shows that the two profiles are similar. Figure 521. Normalized peak voltage variation for Nd:YAG laser; Re= 500; Fr=515; nozzle diameter = 1/ for a mixture of helium and nitrogen to simulate the optical properties of nitrogen. Standard Deviation Profiles. vi 5 *a H imm mmmmmg ;M.AMim... 1.47 1.10 0.74 0.37 0.00 0.37 0.74 1.10 1.47 r/z Figure 522. Percent standard deviation variation for Nd:YAG laser; Re= 500; Fr=515; nozzle diameter = 1/ for a mixture of helium and nitrogen to simulate the optical properties of nitrogen The standard deviation profiles of the mixture for both downstream distances of 2 and 4 nozzle diameters are shown in figure 522. The standard deviation for the mixture Sz/d=2 m z/d=4 0 4 normalized > / concentration 2 1 0 1 2 3 15 10 Sz/d=2 * z/d=4 is more than the standard deviation of pure helium for the two distances. The presence of two flow fluids (helium and nitrogen) shows in the increased standard deviation. Measurements in Backscatter The principal issue in backscatter is distinguishing between the scattered signal and beam dump glare. Figure 523 is a schematic of the backscatter geometry represented on a time scale. The time difference between the scattered beam from the nozzle and reflected beam from the beam dump is approximately 20ns. Thus the photomultiplier tube should be able to distinguish between the two signals 20 ns apart. The minimum bandwidth of the photomultiplier tube required is 5 VMHz. It was impossible to distinguish between the two signals using the Hammamatsu tube. (Bandwidth 23 KHz). 30.4 ns 20.32 ns Nd:YAG laser nozzle Beam dump Photomultiplier Figure 523. Schematic of backscatter on a time basis shows separation between beam dump glare and scattered signal from the nozzle. The waveform seen on the oscilloscope contained both the beam dump glare and the scattered signal. In order to distinguish between the signal and glare the measurements with helium for a typical nozzle position was normalized with a measurement without helium for the same nozzle position. Since the measurements were done in backscatter, the nozzle was traversed in a direction perpendicular to that of the beam(ydirection). Figure 524 shows the data in backscatter. 1.05 c 1u 1 z/d=2 S0.95 E a z/d=6 E 0.9 z/d=10 S0.85 z/d=16 0.8 S0.75 15 10 5 0 5 10 5 10 15 r/z Figure 524. Normalized area variation shows a reduction in scattered intensity in presence of helium (backscatter); Nozzle diameter = 4"; Re =500; Fr = 290 There is a fall in the normalized area in the presence of helium near the jet centerline for all downstream distances. For the case of downstream distance of 2 nozzle diameters, the fall in normalized area is about 0.15 V (15%). For the 900 scattering scheme for the pulsed Nd:YAG laser, the fall in voltage is approximately 30% (Figure 5 16) and 50% for the ArgonIon laser for the downstream case of 2 nozzle diameters for pure helium (Figure 512). The control volume for the backscatter geometry is defend by the 6mm diameter beam and the pulse width of 5 ns (1.5m) (4.29E5m3). For the 900 scattering scheme, the control volume is defined by the 6mm diameter beam and the 0.015mm optical slit (4.29Em3) for the Nd:YAG laser, and 1mm diameter beam and the 0.15mm optical slit for the ArgonIon laser. (7.05E12m3). The leak diameter for the pulse laser is /4"whereas for the ArgonIon laser its /2" Thus for the 900 scattering geometry for both ArgonIon and Nd:YAG cases, the leak occupies 100% of the control volume whereas for the backscatter geometry the leak occupies only 0.6% of the control volume as seen from table 53. Table 53. Control volume to leak diameter ratio for all three cases shows that the ratio is far less for backscatter than 900 scattering. Geometry Nozzle Diameter (D) Control Volume(CV) CV/D (percent) Backscatter,Nd:YAG 6.35E3 m 4.29E5m3 0.6 90 Nd:YAG 6.35E3 m 4.29E5m3 100 90 ArgonIon 12.7E3 m 4.29E5m3 100 This might have lead to the lesser fall in intensity of scattered light in the presence of helium for backscatter geometry. CHAPTER 6 CONCLUSIONS The primary objective of this study is to establish the feasibility of laser induced light scattering as a leak detection technique for hydrogen. There are two primary reasons that support this objective. 1. The fall in voltage at the centerline of the leak indicating reduced scattered intensity both in the presence of pure helium (oHelium = 0.015 GAir) and a mixture of 20% helium and 80% nitrogen (GMixture = GHydrogen= 0.23 GAir) 2. Standard deviation in excess of 30% for pure helium (argon ion laser). Both of these were due to the reduced intensity of Rayleigh scattering by helium molecules. The theoretical studies of Mie scattered intensity show that the Mie signal is the same order of magnitude for the laboratory and maritime particle distributions. This result is important for field measurements as the effect of Mie scattering on the total signal would be approximately the same. Initially it was thought that the fall in voltage would be amplified due to the Mie scatters. However, from the theoretical study it was seen that Mie signal is less than the Rayleigh signal. Hence the lack of Mie scatters in the control volume reduces the fall in voltage in the presence of Rayleigh scattering due to helium and nitrogen molecules. Concentration measurements in the presence of Mie scattering have been successfully done in the lab environment.The backscattering scheme is used principally to test the feasibility of this technique for field measurements. Since for laboratory measurements the scattered signal from the photomultiplier tube and the reflection from the beam dump are only 40 ns apart, it was impossible to distinguish between the two signals for the photomultiplier tube used for the experiments. This lack of temporal discrimination severely limits the overall signal to noise ratio for this configuration. The signal to noise ratio can be improved with better overall temporal response devices. A fall in voltage at the centerline of the flow is observed for both pure helium and the mixture of 20% helium and 80% nitrogen for all four downstream distances. This indicates that a leak could be detected at downstream distances as low as 8 nozzle diameters. Also leak detection is feasible when taken in context of the overall full field concentration distribution. Future Work * In order to determine the effect of background light, a dual line detection system should be used. * A study to determine the polarization effect of the incident beam on the scattered signal should be done. LIST OF REFERENCES Becker H, Hottel H, Williams G, Light scatter technique for the study of turbulent mixing, Journal of Fluid Mechanics, Nov 1967, vol. 30 (2), pp 259284 Bohren C, Huffman D, 1983, Absorption and scattering of light by small particles, Wiley, New York. Bryner N, Pitts W, A Rayleigh Light scattering technique for investigation of free jets and plumes, Review of Scientific Instruments, 1992, vol. 63 (7), pp 36292635 Chen C, Rodi W, 1980, HMT the science and applications of heat and mass transfer, vol. 4, Pergamon Press, Oxford. Dave J, 1970, Scattering of electromagnetic radiation by a large, absorbing sphere. IBM J. Res. Dev. 1969, vol. 13(3), pp 302313. Dibble R, Hollenbach R, Rambach G, Temperature measurements in turbulent flames via Rayleigh scattering, 1980, American Chemical Society, pp 435441. Dyer T, Rayleigh scattering measurements of time resolved concentration in a turbulent propane jet, 1979, AIAA Journal, vol. 17(8), pp 912914 Graham S, Grant A, Jones J, Transient molecular concentration measurements in turbulent flows using Rayleigh light scattering, 1974, AIAA Journal, vol. 12(8), pp 11401142 Horton J, Peterson J, Transient temperature measurements in an ideal gas by using laser induced Rayleigh light scattering, August 1999, Review of Scientific Instruments, vol. 70(8), pp 32223226. Kerker M,1969, The scattering of light and other electromagnetic radiation, Academic Press, New York. Long M, Chu B, Chang R, Instantaneous two dimensional gas concentration measurements by light scattering, AIAA Journal, September 1981, vol. 19(9), pp 11511157. Matthew A, Peterson J, Flow visualizations and transient temperature measurements in an axisymmetric impinging jet rapid thermal chemical vapor deposition reactor, June 2002, Journal of Heat Transfer, vol. 124, pp 564570 McCartney E, 1976, Optics of the atmosphere, Wiley, New York. MullerDethlefs K, Weinberg F, Burning velocity measurements based on laser Rayleigh scattering ,1979, Seventeenth symposium on combustion, vol. 27(2), pp 985992. Otugen M, 1997, Nd:YAG laser based dual line Rayleigh scattering system, AIAA Journal vol. 35(5), pp 776780. Prahl S, http://omlc.ogi.edu/calc/mie_calc.html Last accessed :10/20/2004, Oregon Medical Laser Center. Pitts W, Kashiwagi, 1983, The application of laser induced Rayleigh light scattering to the study of turbulent mixing, Journal of Fluid Mechanics, vol. 141, pp 391429. Pitz R, Cattolica R, Robben F, Talbot L, Temperature and Density in a Hydrogen Air Flame from Rayleigh scattering, 1976, Com. Flame, vol. 27(3), pp 313320. Robben F, Comparison of density and temperature using Raman scattering and Rayleigh scattering using combustion measurements in jet propulsion systems, 1975, Proceedings of a project SQUID workshop, Purdue University. pp 179195. Rodi W, 1982, HMT the science and applications of heat and mass transfer, Pergamon Press, Oxford Rosenweig R, Hottel H, Williams G, Smoke scattered light measurements of turbulent concentration fluctuations, Chem. Eng. Science, 1961, vol. 15(1,2), pp 111129. Schlichting H, 1979, Boundry layer theory, McGraw Hill Series in Mechanical Engineering, New York Shaughnessy E, Morton J, Laser light scattering measurements of particle concentration in a turbulent jet. Journal of Fluid Mechanics, April 1977, vol. 80(1), pp129148. Wiscombe W, 1989, Improved Mie scattering algorithms, Applied Optics, 1980, vol. 19(9), 15051509 BIOGRAPHICAL SKETCH Sameer Paranjpe finished his undergraduate degree in mechanical engineering from the University of Bombay in 2002. He is pursuing his master's degree in mechanical engineering at the University of Florida. He has been a research assistant under Dr. Jill Peterson since August 2002. 