<%BANNER%>

Measurement, Analysis, and Simulation of Wind Driven Rain

Permanent Link: http://ufdc.ufl.edu/UFE0043627/00001

Material Information

Title: Measurement, Analysis, and Simulation of Wind Driven Rain
Physical Description: 1 online resource (236 p.)
Language: english
Creator: Lopez, Carlos Rodolfo
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: driving -- rain -- raindrop -- wind -- wind-driven
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre: Civil Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: This study presents a new experimental approach to collect wind-driven rain data to overcome known issues with data collection in strong winds. Simultaneous measurements of wind and wind-driven rain were collected using a novel tracking system that reorients the rain sensor to maintain correct alignment with the rain trajectory. Experiments were conducted in multiple supercell thunderstorms during the Verification of the Origins of Rotation in Tornadoes Experiment 2 (VORTEX2) and Hurricanes Ike and Irene, which made landfall in the Greater Houston Area in 2008 and the Outer Banks of North Carolina in 2011, respectively. The results of the data analysis appear promising for the continued use of the system and others based on its configuration. In the final component of the study, a wind-driven rain simulation system was designed and implemented at a full-scale test facility for the purpose of evaluating the water penetration resistance of low-rise structures.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Carlos Rodolfo Lopez.
Thesis: Thesis (Ph.D.)--University of Florida, 2011.
Local: Adviser: Masters, Forrest.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043627:00001

Permanent Link: http://ufdc.ufl.edu/UFE0043627/00001

Material Information

Title: Measurement, Analysis, and Simulation of Wind Driven Rain
Physical Description: 1 online resource (236 p.)
Language: english
Creator: Lopez, Carlos Rodolfo
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: driving -- rain -- raindrop -- wind -- wind-driven
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre: Civil Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: This study presents a new experimental approach to collect wind-driven rain data to overcome known issues with data collection in strong winds. Simultaneous measurements of wind and wind-driven rain were collected using a novel tracking system that reorients the rain sensor to maintain correct alignment with the rain trajectory. Experiments were conducted in multiple supercell thunderstorms during the Verification of the Origins of Rotation in Tornadoes Experiment 2 (VORTEX2) and Hurricanes Ike and Irene, which made landfall in the Greater Houston Area in 2008 and the Outer Banks of North Carolina in 2011, respectively. The results of the data analysis appear promising for the continued use of the system and others based on its configuration. In the final component of the study, a wind-driven rain simulation system was designed and implemented at a full-scale test facility for the purpose of evaluating the water penetration resistance of low-rise structures.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Carlos Rodolfo Lopez.
Thesis: Thesis (Ph.D.)--University of Florida, 2011.
Local: Adviser: Masters, Forrest.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043627:00001


This item has the following downloads:


Full Text

PAGE 1

1 MEASUREMENT, ANALYSIS, AND SIMULATION OF WIND DRIVEN RAIN By CARLOS R. LOPEZ A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2011

PAGE 2

2 2011 Carlos R. Lopez

PAGE 3

3 To my parents, Amanda Duarte and Alfonso Lopez

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank my advisor, Forrest J. Masters, Ph.D., P.E. and committee members Kurtis R. Gurle y, Ph.D., David O. Prevatt, Ph.D., P.E. Peter N. Adams, Ph.D., and Katja Friedrich, Ph.D. for their guidance advice, and support throughout my grad uate career I would also like to extend my appreciation to George Fernandez, Jason Smith, J ames Austin, J uan Balderrama, Scott Bolton, Jimmy Jesteadt, Dany Romero, Abraham Alende, and Alon Krauthammer for their assistance in my experiments. This research was made possible by the financial support of the Insurance Institute for Business & Home Safety, National Science Foundation under grants ATM 0910424 (Friedrich) and AGS 0969172 (Friedrich) and the University of Florida Alumni Fellowship Program.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ...... 4 LIST OF TABLES ................................ ................................ ................................ ................ 9 LIST OF FIGURES ................................ ................................ ................................ ............ 10 ABSTRACT ................................ ................................ ................................ ........................ 14 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ ........ 15 Scope of Research ................................ ................................ ................................ ..... 17 Summary of Research Thrusts ................................ ................................ .................. 18 Thrust 1: Development of a Portable Weather Observing System to Characterize Raindrop Size and Velocity in Strong Winds ............................. 18 Thrust 2: Characterization of the Raindrop Size Distribution in Atlantic Tropical Cyclones and Supercell Thunderstorms in the Midwest U.S. ........... 20 Thrust 3: Development of a Wind Driven Rain Simulator for Implementation In a Full Scale Test Facility Capable of Subjecting a Low Rise Building to Windstorm Conditions ................................ ................................ ...................... 21 Importance of the Study ................................ ................................ ............................. 22 Organization of Document ................................ ................................ .......................... 23 2 LITERATURE REVIEW ................................ ................................ .............................. 25 Precipitation ................................ ................................ ................................ ................ 25 Precipitation Events ................................ ................................ ............................. 25 Precipitation Types ................................ ................................ ............................... 26 Raindrop Size Distribution ................................ ................................ ................... 26 Rainfall and Wind Driven Rain ................................ ................................ ................... 30 Quantification of the Rain Deposition on the Building Faade ........................... 31 Factors Affecting Rain Deposition Rate on the Building Faade ....................... 32 Rainfall intensity ................................ ................................ ............................ 32 Influence of the wind on raindrop trajectory ................................ ................. 33 Terrain cha racteristics ................................ ................................ ................... 36 Building characteristics ................................ ................................ ................. 37 Modeling of Rain Deposition on the Building Faade ................................ ............... 38 Semi Empirical Models ................................ ................................ ........................ 38 Numerical Models ................................ ................................ ................................ 40 Full Scale Experiments ................................ ................................ ........................ 41 Measurement of Wind Driven Rain ................................ ................................ ............ 43 Instrumentation ................................ ................................ ................................ .... 43 Limitations of optical disdrometers ................................ ............................... 45

PAGE 6

6 Instrumentation Used In This Study ................................ ................................ .... 47 OTT PARSIVEL optical disdrometer ................................ ............................ 47 Drople t Measurement Technologies Precipitation Imaging Probe .............. 51 Summary ................................ ................................ ................................ ..................... 52 3 DESIGN, PROTOTYPING, AND IMPLEMENTATION OF ARTICULATING RAIN PARTICLE SIZE MEASUREMENT PLATFORMS ................................ .......... 53 Motivation for the Development of an Articulating Instrument Platform ................... 53 Raindrop Size Distri bution Verification Using the Oil Medium Test ................... 54 Bearing Drop Test ................................ ................................ ................................ 56 Instrument Performance in High Wind Speeds ................................ ................... 60 Articulating Instrumentation Platforms ................................ ................................ ....... 64 Articulating Instrumentation Platform Components ................................ ............ 64 RM Young sonic anemometer ................................ ................................ ...... 65 Articulating support structure ................................ ................................ ........ 66 Data acquisition system ................................ ................................ ................ 67 Power ................................ ................................ ................................ ............. 68 Substructure ................................ ................................ ................................ .. 68 Portable Instrument Platform Operation ................................ .............................. 69 Quality Control Algorithm ................................ ................................ ..................... 70 Description of Weather Station T 3 ................................ ................................ ..... 71 Comparison of Collocated Stationary and Articulating Instruments in a Supercell Thunderstorm ................................ ................................ ................................ .......... 73 Summary ................................ ................................ ................................ ..................... 75 4 CHARACTERIZATION OF WIND DRIVEN RAIN IN STRONG WINDS .................. 76 Field Research Programs ................................ ................................ ........................... 76 Verification of the Origins of Rotation in Tornadoes Experiment 2 (VORTEX2) ................................ ................................ ................................ ....... 76 Overview ................................ ................................ ................................ ........ 76 Deployment details ................................ ................................ ........................ 77 Florida Coastal Monitoring Program (FCMP) ................................ ...................... 79 Overview ................................ ................................ ................................ ........ 79 Deployment details ................................ ................................ ........................ 80 Effect of Wind Velocity and Turbulence Intensity on Raindrop Diameter ................. 82 Wind Velocity and Turbulence Intensity Dependency of the Raindrop Size Distribution ................................ ................................ ................................ ............... 90 Comparison of Raindrop Size Dist ribution Models to Measured Raindrop Size Distribution Data in Multiple Wind Velocities ................................ ........... 95 Peak to Mean Ratio of Rainfall Intensities ................................ ............................... 101 Comparison of Ground Measured Rainfall Intensity and Estimated Reflectivity to Weather Surveillance WSR 88D Estimated Rainfall Intensity and Measured Reflectivity ................................ ................................ ................................ ............. 105 Summary ................................ ................................ ................................ ................... 108

PAGE 7

7 5 DEVELOPMENT OF A WIND DRIVEN RAIN SIMULATION SYSTEM FOR THE WATER PENETRATION RESISTANCE EVALUATION OF LOW RISE BUILDINGS IN A FULL SCALE WIND TUNNEL ................................ .................... 109 Design Specifications for the Rain Simulator in the IBHS Research Center .......... 110 Spray Uniformity and the Effect of Wind Velocity on the Raindrop Size Distribution ................................ ................................ ................................ ............. 112 Characterization of the Raindrop Size Distribution of a Spay Nozzles in Stagnant Air: A Proxy for Full Scale Testing ................................ ........................ 115 Specimen Matrix ................................ ................................ ................................ 115 Experimental Configuration ................................ ................................ ............... 116 Analysis ................................ ................................ ................................ .............. 116 Validation of the Wind Driven Ra in Simulation System at the IBHS Research Facility ................................ ................................ ................................ .................... 122 Summary ................................ ................................ ................................ ................... 125 6 SUMMARY, CONCLUSIONS,AND RECOMMENDATIONS ................................ .. 126 Characterization of Extreme Wind Driven Rain Events ................................ .......... 126 Proof of Concept ................................ ................................ ................................ 126 Conclusions from Field Data Results ................................ ................................ 128 Effect of wind velocity and turbulence intensity on raindrop diameter ...... 128 wind velocity and turbulence intensity dependency of the raindrop size distribution ................................ ................................ ................................ 129 Comparison of Measured Data to Raindrop Size Distribution Models ............. 129 Comp arison of Measured Data to WSR 88D Data ................................ ........... 130 Peak to Mean Ratio of Rainfall Intensities ................................ ........................ 130 Design and Implementation of a Full Scale Wind Driven Rain System .................. 131 Nozzle Characterization ................................ ................................ ..................... 131 Full Scale Implementation ................................ ................................ ................. 131 Recommendations for Future Research ................................ ................................ .. 132 Recommendations for Instrumentation ................................ ............................. 132 Recommendations for Full Scale and Numerical Models ................................ 133 Recommendations for the Morphological Image Processing Algorithm .......... 134 APPENDIX : ADDITIONAL INFORMATION AND DATA ................................ ................ 135 Theoretical Proof of Greater Accuracy from an Articulating Instrument ................. 135 Nozzle Selection ................................ ................................ ................................ ....... 139 Measured Diameter Wind Relationships ................................ .............................. 141 Hurricane Ike Data ................................ ................................ ............................. 141 Hurricane Irene (Beaumont) Data ................................ ................................ ..... 144 Hurricane Irene (Deal) Data ................................ ................................ ............... 147 Florida Coastal Monitoring Program Hurricane Ike Data ................................ ........ 150 Verification of the Origins of Rotation in Tornadoes Experiment 2 Articulating Instrument Platform Data ................................ ................................ ...................... 153 Measured Nozzle Characteristics ................................ ................................ ............ 210

PAGE 8

8 Uniformity ................................ ................................ ................................ ........... 210 Measured Nozzle Raindrop Size Distribution ................................ ................... 219 REFERENCES ................................ ................................ ................................ ................ 224 BIOGRAPHICAL SKETCH ................................ ................................ .............................. 235

PAGE 9

9 LIST OF TABLES Table page 2 1 PARSIVEL diameter classes ................................ ................................ ................. 48 2 2 PARSIVEL velocity classes ................................ ................................ ................... 49 3 1 Trajectory angles of multiple diameter drops in multiple wind speeds ................. 57 3 2 Steady wind test matrix ................................ ................................ .......................... 61 4 1 Verification of the Origins of Rotation in Tornadoes Experiment 2 deployment details ................................ ................................ ................................ ...................... 78 4 2 Peak to mean ratios of U and R ................................ ................................ ........... 102 4 3 Comparison of Z R models ................................ ................................ .................. 106 5 1 Rainfall Intensities ................................ ................................ ............................... 111 5 2 Spray nozzles evaluated in this study ................................ ................................ 115

PAGE 10

10 LIST OF FIGURES Figure page 1 1 Articulating precipitation m easurement platform and stationary platform ............ 19 1 2 Portable FCMP weather station with the actively controlled positioning system ................................ ................................ ................................ .................... 20 1 3 Insurance Institute for Business & Home Safety Research Center ...................... 21 2 1 Influence of the modified three parameter gamma model parameters ................ 29 2 2 Components of the rain intensity vector ................................ ................................ 31 2 3 Drop size and shape ................................ ................................ .............................. 33 2 4 Trajectory of drops released at 0 m/s ................................ ................................ .... 35 2 5 Distance required for the traj ectory angle to reach 95% of the terminal angle .... 36 2 6 Deposition of smaller drops left and larger drops right ................................ ......... 38 2 7 PARSIVEL drop diameter and velocity determination ................................ .......... 50 2 8 PARSIVEL measurement and output ................................ ................................ .... 50 3 1 Sample picture from morphological image processing algorithm ......................... 55 3 2 Comparison of different raindrop size distribution measurement techniques ...... 56 3 3 Angled trajectory experiment configuration ................................ ........................... 57 3 4 PARSIVEL measured diameters at multiple trajectory angles ............................. 58 3 5 PARSIVEL measured velocities at multiple trajectory angles .............................. 59 3 6 Steady wind instrument configuration ................................ ................................ ... 61 3 7 PARSIVEL measured raindrop size distributions for the tests in steady wind flow ................................ ................................ ................................ .......................... 62 3 8 PARSIVEL measured drop diameters and velocities for the tests in steady wind flow ................................ ................................ ................................ ................. 63 3 9 Articulating instrument platform ................................ ................................ ............. 65 3 10 Articulating instrument platform automation ................................ .......................... 66

PAGE 11

11 3 11 Drag coefficient of a cylinder dependant of Reynolds number of flow ................. 67 3 12 Instrument platform control and data diagram ................................ ...................... 68 3 13 Quality control filter ................................ ................................ ................................ 71 3 14 T 3 Precipitation Imaging Probe turret syst em a nd Gill anemometers at ............ 72 3 15 Measured raindrop size distribution by stationary instrumentations and articulating instrumentation ................................ ................................ .................... 74 3 16 Comparis on of rainfall intensities measured by stationary and articulating instrument platforms ................................ ................................ ............................... 74 3 17 Comparison of estimated reflectivity by stationary and articulating instrument platforms ................................ ................................ ................................ ................. 75 4 1 VORTEX2 instrument deployment ................................ ................................ ........ 79 4 2 VORTEX2 data collection sites ................................ ................................ .............. 79 4 3 Effect of longitudinal wind velocity on drop diameter observed in VORTEX2 data ................................ ................................ ................................ ......................... 84 4 4 Effect of longitudinal turbulence intensity on drop diam eter observed in VORTEX2 data ................................ ................................ ................................ ....... 85 4 5 Effect of lateral turbulence intensity on drop diameter observed in VORTEX2 data ................................ ................................ ................................ ......................... 86 4 6 Effect of longitudinal wind velocity on drop diameter observed in FCMP data .... 87 4 7 Effect of longitudinal turbulence intensity on drop diameter observed in FCMP data ................................ ................................ ................................ ............. 88 4 8 Effect of lateral turbulence intensity on drop diameter observed in FCMP data .. 89 4 9 VORTEX2 gamma parameters observed in multiple wind conditions and rainfall intensities ................................ ................................ ................................ .... 91 4 10 FCMP Hurricane Ike and Irene gamma parameters observed in multiple wind conditions and rainfall intensities ................................ ................................ ........... 92 4 11 FCMP and VORTEX2 gamma parameters observed in multiple wind conditions and rainfall intensities ................................ ................................ ........... 93 4 12 Observ ed shape slope relation ................................ ................................ .............. 94

PAGE 12

12 4 13 Model raindrop size distribution and measured raindrop size distribution comparisons ................................ ................................ ................................ ........... 96 4 14 Model raindrop size distribution and measured raindrop size distribution comparisons ................................ ................................ ................................ ........... 97 4 15 Model raindrop size distribution and measured raindrop size distribution comparisons ................................ ................................ ................................ ........... 98 4 16 Mean square error values of raindr op size distribution models stratified by U and R ................................ ................................ ................................ ...................... 99 4 17 Mean square error values of raindrop size distribution models stratified by TI U and R ................................ ................................ ................................ ...................... 99 4 18 Mean square error values of raindrop size distribution models stratified by TI V and R ................................ ................................ ................................ .................... 100 4 19 Peak to mean ratios of VORTEX2 data ................................ ............................... 103 4 20 Peak to mean ratios of FCMP data ................................ ................................ ..... 104 4 21 Comparison of disdrometer estimated and radar measured reflectivity ............. 106 4 22 Comparison of disdrometer measured and radar estimated rainfall intensity .... 107 4 23 Observed Z R relationship ................................ ................................ ................... 107 4 24 Comparison of observed and recommended Z R relationships ......................... 108 5 1 Apparatus to measure the spray uniformity of a single nozzle ........................... 113 5 2 Comparison of raindrop size distributi ons measured in stagnant air and in a steady wind by the PARSIVEL and PIP ................................ .............................. 115 5 3 PARSIVEL test locations for nozzle characterization ................................ ......... 116 5 4 Count of drops radially outward from nozzle centerline ................................ ...... 119 5 5 Determining initial velocity using high speed foot age ................................ ......... 120 5 6 Raindrop size distribution of BETE WL3 in stagnant air conditions ................... 121 5 7 Instrument arrangement at the Insurance Institute for Business & Home Safety Research Center ................................ ................................ ....................... 123 5 8 Measured raindrop size distributions at multiple height s and wind velocities .... 124 5 9 Comparison of measured raindrop size distributions and Best (1950) model ... 125

PAGE 13

13 A 1 D roplet formation process ................................ ................................ .................... 139 A 2 Types of nozzles and drop size relationship ................................ ....................... 140

PAGE 14

14 Abstract of Dissertation Presented to the Graduate School of the Uni versity of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy MEASUREMENT, ANALYSIS, AND SIMULATION OF WIND DRIVEN RAIN By Carlos R. Lopez December 2011 Chair: Forrest J. Masters Major: Civil Engine ering This study presents a new experimental approach to collect wind driven rain data and overcome known issues associated with field measurements during strong winds. Simultaneous measurements of wind and wind driven rain were collected using a novel tr acking system that continuously reorients a raindrop size spectrometer (or disdrometer) to maintain correct alignment with the rain trajectory. Experiments were conducted in multiple supercell thunderstorms during the Verification of the Origins of Rotatio n in Tornadoes Experiment 2 (VORTEX2), and Hurricanes Ike (2008) and Irene (2011) with the Florida Coastal Monitoring Program (FCMP fcmp.ce.ufl.edu ). The results of the data analysis appear promising for the continued use of the system and others based on the same concept. The final component of the study consisted of the design and implementation of a wind driven rain simulation system at a full scale test facility to evaluate the water penetration resistance of low rise structures.

PAGE 15

15 CHAPTER 1 INTRODUCTION This s tudy presents a new experimental approach to collect wind driven rain data and overcome known issues associated with field measurements during strong winds. Simultaneous measurements of wind and wind driven rain were collected using a novel tracking system that continuously reorients a raindrop size spectrometer (or disdrometer) to maintain correct alignment with the rain trajectory. Experiments were conducted in multiple supercell thunderstorms during the Verification of the Origins of Rotation in Tornadoe s Experiment 2 (VORTEX2), and Hurricanes Ike (2008) and Irene (2011) with the Florida Coastal Monitoring Program (FCMP fcmp.ce.ufl.edu ). The results of the data analysis appear promising for the continued use of th is system and others based on the same co ncept. The final component of the study consisted of the design and implementation of a wind driven rain simulation system at a full scale test facility to evaluate the water penetration resistance of low rise structures. The motivation for this research i s the extensive damage caused by tropical cyclones annually. I n the last two decades, Atlantic tropical cyclones have caused more than $113 billion (2009 dollars) in insured losses (Insurance Information Institute, 2011) Post storm investigations (e.g., Mehta et al.,1983; NIST, 2005; FEMA, 2005; FEMA, 2006; Guillermo et al., 2010 ; Gurley and Masters, 2011 ) have found that building envelope failures are a leading cause of damage. A critical, recurring problem in residential co nstruc tion is water ingress through the building envelope W ind related failures causing mismanaged or unmanaged water infiltration can result in loss of function and costly damage to building contents ( Lstiburek, 2005; Mullens et al., 2006; IBHS, 2009) W DR deposition on the building

PAGE 16

16 faade can cause moisture accumulation in porous wall components (Bondi and Stefanizzi, 2001; Abuku et al., 2009; Bomberg et al., 2002; Tsongas et al., 1998; Lang et al., 1999) deterioration of wood frame wall systems (Hulya et al., 2004; Lacasse et al., 2003) water infiltration through the secondary water barrier of roof systems (Bitsuamlak et al. 2009) and the water penetration resistance of residential wall systems with integrated fenestration (Salzano et al., 2010; Lopez et al., 2011) Wind driven rain is an active research subject in the atmospheric sci ence, engineering, and building science disciplines E ngineering and building science research has primarily focused on modeling wind driven rain deposition on building faades ( Blocken and Carmeliet, 2010; Blocken and Carmeliet, 2004; Choi, 1994; Choi, 1994; Surry et al., 1994) hygrothermal performance and drying (Teasdale St Hilaire and Derome, 2007; Abuku et al., 2009; Cornick and Dalgliesh, 2009) and to a lesser extent, fragility modeling (Dao and Van de Lindt, 2010) In atmospheric science, extensive research has been directed toward the improve ment of a) radar and satellite derived estimations of rainfall intensity (e.g., Wilson and Brandes, 1979; Rosenfeld et al., 1993 ; Kedem et al., 1994) and b) microphysical models for numerical weather prediction ( Chen and Lamb, 1994; Gaudet and Cotton, 1998; Stoelinga et al., 2003) Raindro p size distributio n (RSD) is a critical variable in both fields. One view of features aloft, while engineering uses the RSD as a probabilistic input for modeling raindrop trajectories and rain deposition rates on buildings. In both applications the relative difference between the number and concentration of small and large drops is also critical. In atmospheric science, remote measurements of the radar reflectivity of

PAGE 17

17 weather systems are partic ularly sensitive to the drop size. In engineering, the wetting pattern and the rain deposition rate on the building faade is in part a function of the RSD. Since the 1940s, a significant amount of RSD data has been collected during stratiform and conv ective precipitation events in many locations around the world. However, in situ wind driven rain data are scarce. The knowledge base is largely built from radar and aircraft observations or adapted from in situ measurements collected in little to no wind. For example, the Best (1950) RSD, which is widely used in computational wind engineering was not calibrated with RSD data collected in high winds yet it is still used to model wind driven rain deposition on the building faade. This research directly add resses this issue through multiple thrusts, which are described in the next section. Scope of Research The study addresses wind driven rain (WDR) in an interdisciplinary context, with emphasis on the characterization of field observations (atmospheric scie nce) and the simulation of a wind driven rain field to evaluate the water penetration resistance of the building envelope (engineering). First, a portable weather observation system was developed to obtain reliable particle size distributions in strong win ds. Second, the system was field tested during an in situ data collection campaign throughout the southern and central Plains as part of the Verification of the Origins of Rotation in Tornadoes Experiment 2 (VORTEX2) D ata obtained during the VORTEX2 campa ign is also compared to measurements collected during Hurricane Ike (2008, separate study) and Hurricane Irene (2011, led by the author). Third, a rain simulation system for a full scale test facility was developed; the design criteria were established fro m the results of

PAGE 18

18 the first and second thrusts. The purpose of this thrust is to assist the Insurance Institute for Business & Home Safety (IBHS) in the commissioning of its full scale test facility to simulate wind driven rain effects. A spray system was designed, installed and tested based on the results obtained from the first and second thrusts. Summary of Research Thrusts Thrust 1: Development of a Portable Weather Observing System to Characterize Raindrop Size and Velocity in Strong Winds The first contribution of this research was to design, prototype and successfully field evaluate a portable weather observation system to quantify hydrometeor size and velocity in strong winds ( Figure 1 1 left). The system operates by conti nuously readjusting the orientation of an OTT PARticleSIze and fall VELocity disdrometer (PARSIVEL) to maintain optimal alignment with the raindrop trajectory. Disdrometers function by measuring the voltage drop from a photodiode, or series of photodiodes, caused by a raindro p passing through a light band (Loffler Mang and Joss, 2000) Reliable rain data acquisition via o ptical disdrometers requires that the hydrometeors travel nearly perpendicular to the light plane. In little to no wind, this presents no issue for a stationary instrument with the light plane parallel to the ground ( Figure 1 1 right) In the presence of strong winds, advection is a dominant component of the particle trajectory. A statio nary instrument loses accuracy as the wind speed increases. Thus an outstanding experimental challenge has been the development and implementation of an observational system capable of accurately quantifying the RSD during an extreme wind event While act ively aligning the disdrometer with the mean rain vector was previously considered by Grifftihs (1974) this research presents the first such known effort to successfully address this issue. Two systems using different instruments

PAGE 19

19 were implemented. The fir st system employs a Droplet Measurement Technologies Precipitation Imaging Probe (DMT PIP), while the second utilizes a PARSIVEL disdrometer. 1. PARSIVEL based systems: A PARSIVEL disdrometer and a sonic anemometer were installed on articulating platforms wit h actively controlled positioning systems. Figure 1 1 depicts the PARSIVEL disdrometers mounted on articulating and stationary platforms; both platforms were deployed repeatedly throughout a six week field campaign in the southern and central Plains during 2010 for the VORTEX2 experiment. In 2011 both systems (the PIP and PARSIVEL based systems) were deployed during the passage of Hurricane Irene. 2. PIP based system: A DMT PIP was installed on an actively controlled m echanized turret on a Florida Coastal Monitoring Program (FCMP) weather station ( Figure 1 2 ) Th is instrument was first deploy ed successfully during Hurricane Ike (2008) Figure 1 1 Articulating precipitation m easurement platform ( left) and stationary platform ( right p hoto courtesy of author )

PAGE 20

20 Figure 1 2 Portable FCMP weather station with the actively controlled positioning system (photo courtesy of author) Thrust 2: Characterization of the R aindrop S ize D istribution in Atlantic Tropical Cyclones and Supercell T hunderstorms in the Midwest U.S. With the use of the new instrument platforms, multiple measurements were taken in supercel l thunderstorms and in tropical cyclones as part of the VORTEX2 and FCMP field campaigns. e, are the first set of RSD measurements in high winds. Thus, the a nalysis and results of the se data are presented in this docume nt to address the following questions in an effort to advance the WDR knowledge base : 1. D oes wind velocity and turbulence intensity affect rainfall characteristics (e.g. mean drop diameter, RSD, etc.)? 2. Are existing RSD models based on data collected in littl e to no wind applicable to WDR occurring in an extreme wind events? 3. What is the relationship between the peak short duration rainfall intensity to a long term average and how does this affect current WDR specifications?

PAGE 21

21 4. Does the precipitation algorithm i n the radar product generator used by the National Weather Service (NWS) Weather Surveillance Radar exhibit any biases that manifest in extreme wind events? 5. Based on the results of 1 4 what are the implications for computational and experimental simulatio n and specification of design requirements for water penetration resistance of building products and systems? Thrust 3: Development of a W ind D riven R ain Simulator for Implementation In a Full Scale Test Facility Capable of Subjecting a Low Rise Building t o Windstorm Conditions In addition to answering the preceding questions, the data was also used to develop a metho d to design a WDR simulation system to be used by the Insurance Institute for Business & Home Safety (IBHS) IBHS recently constructed a 30 MW full scale test facility that can replicate windstorm conditions at a sufficient scale to evaluate the performance of a two story building subjected to hurricane conditions ( Figure 1 3 ). The results obtained from the field resear ch were used to determine a realistic RSD for the simulations. Tests were then performed to determine the effectiveness of the PARSIVEL as a reference instrument, and once the results were verified, a range of commonly available nozzles were evaluated to d etermine the optimal choice for the facility. Figure 1 3 I nsurance Institute for Business & Home Safety Research Center (photo courtesy of IBHS) IBHS.ORG

PAGE 22

22 Importance of the Study The research findings have the pot ential to impact atmospheric sciences and wind engineering. In atmospheric research, calibrating radar precipitation estimate algorithms to ground level rain gauges has led to better predictions and estimations of rainfall intensity and accumulation (Habib et al., 2009) Few such gauges exist (Linsley et al., 1992) and most are not designed to operate in high winds. In contrast, this study deployed multiple instrument stations in a user defined arra y. These data can be used to select rainfall reflectivity and rainfall intensity (Z R) relationships for extreme wind event precipitation systems. Furthermore, the ground level data will complement cloud level radar data and could be used to model cloud to ground RSD evolution (Wilson and Brandes, 1979) In c omputational wind engineering research, t o model WDR trajectories current analysis techniques employ the following aspects: (1) the wind flow pattern is calculated by solvi ng the three dimensional Reynolds Averaged Navier Stokes equations, the continuity equation, and the equations of the realizable k turbulence model; (2) a choice of drag coefficient formulas for drops or experimentally derived drag coefficients from Gunn and Kinzer (1949); (3) the use of models to simulate turbulence dispersion of drops; and (4) a spatially and temporally constant RSD, modeled after work performed by Best (1950). Choi (1994) Blocken and Carmeliet (2002) among others (Rodgers et al., 1974; Inculet and Surry, 1994; Nore et al., 2007) have shown that drop size affects the trajectory of particles near a bluff body. Thus, the data collected were compared with established models (e.g., Marshall and Palme r 1948, Best 1950 Willis and Tattleman 1989 and the three parameter gamma model using the mean of the parameters calculated from the gathered data ) to determine their appropriateness for

PAGE 23

23 simulating extreme WDR conditions This information will be made ava ilable to the American Society of Civil (ASCE) Task Committee on Wind Driven Rain Effects which is currently preparing a state of the art report for the ASCE Technical Council on Wind Engineering. Experimental wind engineering will also benefit from new data gathered in high wind by setting more comprehensive simulation requirements. This is an emerging sub discipline and o nly a few WDR studies have been conducted at full scale (e.g., Salzano 2010, Bitsuamlak 2009) and in the wind tunnel (e.g., Inculet 1994) In the full scale experiments, the primary focus was replicating dynamic wind loading and faade wetting rates determined from current test standards (e.g., ASTM E331 00, ASTM E547 00, ASTM 1105 05); therefore, correct RSD simulation was not a concern. The RSD prescribed in the wind tunnel experiments was determined from models (Best, 1950; Marshall and Palmer, 1948) based on data collected in little to no wind. In the design of the IBHS WDR simulation sy stem, the collected data was used to determine the required RSD. This methodology can ultimately lead to improved performance evaluation of the water penetration resistance of building products and systems. Organization of Document Chapter 2 provides an o verview of precipitation, the definition of WDR, factors that influence WDR, different WDR measurement techniques, and a review of previous computational and experimental simulation methodologies. Chapter 3 explains the design, prototyping, and field evalu ation of the articulating instrumentation platform. Chapter 4 presents the characterization of WDR in strong winds. Chapter 5 discusses the technical approach that was taken to design a full scale WDR system, and Chapter

PAGE 24

24 6 summarizes the research efforts a nd provides conclusions and recommendations for future research.

PAGE 25

25 CHAPTER 2 LITERATURE REVIEW This chapter presents fundamental concepts of precipitation and wind driven rain (WDR). A description of the factors that influence the deposition of rain on the buildi ng faade is also provided. Finally, raindrop size distribution (RSD) measurement techniques and previous computational and experimental simulation methodologies are reviewed. Precipitation Precipitation Events Precipitation results from the cooling of war m air generally occurring as warm air rises over cold air masses. As the air cools, water vapor condenses on particulate matter in the air (e.g. dust and salt particles). Drops then grow by either collision and coalescence (in warm conditions) or ice crys tal growth (i.e. the Bergeron Findeisen process; Houze, 1994). Drop growth continues until the hydrometeors are large enough to overcome updrafts and precipitation occurs as they fall under the influence of gravity. Orographic effects, frontal systems, or convection are some of the processes that cause lifting of air masses in the atmosphere. Orographic lift forces result from the upward deflection of horizontally moving air masses that encounter large orographic obstructions. As the air mass is deflected u pward, it cools allowing condensation and subsequently precipitation. Frontal systems occur when a cold air mass approaches a warm air mass (cold front) or vice versa (warm front). Cold fronts generate precipitation that is usually high intensity, short du ration, and occurs over a limited area. Severe thunderstorms are associated with this type of frontal system. Conversely, warm fronts generate mild, long term, and widespread precipitation. Convection occurs when solar

PAGE 26

26 radiation warms the earth surface and subsequently the adjacent air. As the air mass warms, it rises, and precipitation occurs (Houze, 1994) This is the process that forms convective thunderstorms in the rainbands and eyewall of tropical cyclones (Gray, 1979) Precipitation Types Precipitation is generally classified as stratiform or convective. The difference is attributed to the characteristic vertical air velocity in the cloud structure. In stratiform precipitation, the vertical ai r velocity is much less than the fall velocity of the hydrometeors. In convective precipitation, the vertical air velocity is on the order of the fall velocity of the hydrometeors. As a result, hydrometeors in convective precipitation spend less time airbo rne (Houze, 1994) and are smaller than their stratiform counterparts. Convective precipitation generally produces more intense rainfall rates over a shorter duration than stratiform precipitation (To kay et al., 1999) In addition, convective precipitation can produce a greater concentration of small to medium sized drops and fewer concentrations of large drops than stratiform precipitation (Tokay and Short, 1996) Raindr op Size Distribution The raindrop size distribution (RSD) refers to the concentration of all drop sizes for a given sample volume. Early models of the RSD were developed under the assumption that a reference bulk variable usually horizontal rainfall inte nsity (defined in the next section) is the governing parameter (cf. Torres et al. 1994). Marshall and Palmer (1948) were the first to develop a rainfall dependent RSD ( ) model: ( 2 1 )

PAGE 27

27 ( 2 2 ) ( 2 3 ) where is the intercept parameter, is the slope parameter, is the horiztonal rainfall intensity. The most widely used model in building science and wind engineering is the Best (1950) model: ( 2 4 ) ( 2 5 ) ( 2 6 ) where is the fraction of liquid wat er in the air consisting of raindrops of radii < (mm), (mm/hr) is the horizontal rainfall intensity, and (mm3/ m3) is the volume of liquid water per unit of volume of air. The main difference between the Best (1950) and the Marshall and Palmer (19 48) models is that Best model does not assume a constant intercept parameter and is not linear in log space. Ulbrich (1983) demonstrated that RSDs can vary significantly under different types of rainfall conditions. He proposed a modified three parameter g amma distribution ( 2 7 ) which has become a widely accepted method for fitting RSDs ( 2 7 ) where is the intercept parameter (mm 3 m 1), is the shape parameter (dimensionless), and is the slope parameter (mm 1). Moments ( 2 8 ) of the measured RSD are used to estimate the three parameters ( and ). For this study the M246 moment estimator method is employed (Cao and Zhang, 2009) :

PAGE 28

28 ( 2 8 ) ( 2 9 ) ( 2 10 ) ( 2 11 ) ( 2 12 ) where is the moment order, is the ratio of moments and is the gamma function defined as: ( 2 13 ) Figure 2 1 demonstrates the model sensitivity to each parameter. The dark line depicts RSD derived from the F CMP data acquired during Hurricane Ike. and characterize the empirical distribution shown (dark line). The sensitivity to each parameter is illustrated by independently changing and to the 5th and 95th percentiles of the parameter estimates. The intercept and slope parameters indicate the shift and rotation of the distribution, respectively, and the shape parameter indicates the concavity of th e distribution.

PAGE 29

29 Figure 2 1 Influence of the modified three parameter gamma model parameters Willis and Tattleman (1989) expanded the gamma model by researching the la rge fluctuations associated with high rainfall intensities They developed a method for estimating the three parameters using rainfall intensity as the only reference variable and calibrated to data collected in extreme events (Hudson, 1970; >100mm/hr). T he Wills Tattleman model (1989) uses empirically derived formulas for water content ( ) and median volume diameter ( ). The equations to estimate the three parameters were developed using a fit to the normali zed data:

PAGE 30

30 ( 2 14 ) ( 2 15 ) ( 2 16 ) ( 2 17 ) ( 2 18 ) Validation of the Willis and Tattelman (19 89) model was accomplished via a comparison to approximately 14,000 ten second samples collected from hurricanes and tropical storms from 1975 1982 at 3000 m (9843 ft) and 450 m (1476 ft). The results indicate that the model reasonably characterized the observed distributions collected in rainfall rates of up to 225.0 mm/hr (8.9 in/hr). Rainfall and Wind Driven Rain Horizontal rainfall intensity refers to the accumulated volume of rain caused by the flux of rain through a horizontal plane. Wind driven ra in (WDR) occurs when wind induced drag forces impart a horizontal component of motion to the falling particles. The components of the rain intensity vector ( and ) are defined as follows: = the oblique rain vector = the acc umulated volume of rain, over a specified amount of time, caused by the flux of rain through a horizontal plane. the accumulated volume of rain, over a specified amount of time, caused by the flux of rain through a vertical plane and are illustrated in Figure 2 2 (after Blocken and Carmeliet, 2002).

PAGE 31

31 Figure 2 2 Components of the rain intensity vector Quantification of the Rain D eposition on the Building Faade Methods for quantifying wind induced wetting of the building faade were developed by Choi (1993) Straube and Burnett (2000) and Blocken and Car meliet (2002) The WDR deposition on the building is defined as the ratio of wetting on the building at a specified location on the faade to the driving rain intensity ( ). The terminology used in this document is based on Choi (1994). The Local Effect Factor ( ) is the ratio at time of the WDR intensity ( ) at a particular location on the faade to the unobstructed horizontal rainfall intensity ( ) in the free stream for a single hydrometeor of diameter : ( 2 19 ) The equivalent parameter for the deposition of all raindrop sizes at a location on the faade is the Local Intensity Factor ( ). The is obtained by integrating the s over all hydrometeor diameters (Choi, 1994):

PAGE 32

32 ( 2 20 ) The wett ing of the building faade is highly non uniform. RSD, wind, terrain, and building characteristics are factors that influence the wind flow around the building, the drop trajectories, and consequently, the wetting of the building facade. These effects are now discussed. Factors A ffecting Rain Deposition Rate on the Building Faade Rainfall i ntensity The amount of rain impinging on the building surface is primarily dependent on the amount of precipitation in the boundary layer. This quantity is defined as t he unobstructed horizontal rainfall intensity. Ground measurements of horizontal rainfall intensity have been primarily collected for hydrological and agricultural purposes. The sampling interval of these data is seldom less than three to six hours and mor e commonly between daily and monthly (Willis and Tattelman, 1989) These time scales are inadequate for engineering applications (Blocken and Carmeliet,2005) that require continuous, high resolution time histories. Most test methods for the water penetration resistance of building systems (e.g. ASTM E331 00, ASTM E1105 00, ASTM E2268 04, and ASTM E547 00) prescribe a wetting rate of 203 mm/hr; this quantity reflects the rate required to cause water to sheet over a curtain wall. The National Weather Service (NOAA, 1977) provides 5 to 60 minute precipitation frequency atlases for the eastern and central United States, in which the maximum rainfall intensity for the South E astern United States is 274 mm/hr for a 100 year return 5 min rain event.

PAGE 33

3 3 Influence of the w ind on raindrop t rajectory Rainfall trajectories are influenced by body forces (e.g., gravity) and surface forces (e.g., wind induced drag). Condensation of water vapor on particulate matter in the air during drop synthesis produces small drops nearly spherical due to the surface tension dominating over pressure forces. Throughout their freefall, drops continuously collide, coalesce, and break up yielding different sized drops. Smaller sized drops are susceptible to evaporation while larger drops are subjected to unequal pressure distributions that cause distortion ( Figure 2 3 ). This deviation from the spherical assumption c an lead to over estimation of drag coefficients, particularly at high velocity, high Reynolds number flow (Hu & Srivastava, 1995) Figure 2 3 Drop size and shape Raindrops tr aveling through the boundary layer in the free stream are assumed to have a horizontal velocity component that asympt otically approaches the wind speed due to drag forces and a fall velocity (vertical component) equal to their terminal velocity. Figure 2 4 and Figure 2 5 depict drop trajectories for various drop diameters and wind velocities; the trajectories were modeled assuming a steady flow of marked velocity, and the drop drag coefficients and terminal velocities given by Gunn and Kinzer (1949, model is explained in detail in Chapter 5) Figure 2 5 indicates that the distance at which drops have achieved 95% of the theoretical trajectory angle the

PAGE 34

34 angle at which the horizontal component of the drop velocity is equal to the wind velocity is less than 55 m. Masters et al. (2010) reported a minimum mean and standard deviation values of 126.2 43.0 m, 74.4 31.9 m, and 120.4 54.4 m for 15 minute integral scales (at an elevation of 10 m) during Hurricanes Katrina, Rita, and Wilma in 2005, respectively. Thus, assuming that the drop travels at the gradient wind speed in tropical cyclones is va lid. As raindrops approach the building faade, higher wind speeds increase the horizontal component of motion. With higher horizontal velocities, more drops are susceptible to striking the building surface. Choi (1994) found that changing wind velocity from 5.0 m/s to 30.0 m/s can increase LIF values up to 10 times for the top quarter of a 4:1:1 ratio building. Therefore, increasing wind velocity will increase the effect of all raindrop sizes on the building faade, p articularly in the top quarters. A)

PAGE 35

35 B) C) Figure 2 4 Trajectory of drop s released at 0 m/s A)

PAGE 36

36 B) C) Figure 2 5 Distance required for the trajectory ang le to reach 95% of the terminal angle Terrain characteristics Terrain characteristics affect the wetting of the building faade by changing the characteristics of the flow upwind particularly the mean wind speed and turbulence intensity Karagiozis et al. (1997) described the terrain characteristics affecting the flow conditions upwind of the building faade; these characteristics range from ground surface roughness dictating the global terrain exposures and overall flow conditions (e.g., open, suburban, u rban) to larger obstructions introducing local disturbances to the

PAGE 37

37 upwind flow (e.g., close vicinity building obstruction). Significant distortions of the flow pattern directly upwind of the building resulting from the introduction of high turbulence and m ixing due to blockage and shielding effects of building obstructions at close vicinity causes the distribution of wetting on the building faade to deviate from what is commonly expected (Choi, 1993) When no large obstructions are directly upwind, the greatest effect to the wind flow pattern is due to the aerodynamic roughness length ( ) The aerodynamic roughness length is d efined as the height at which the mean velocity is zero assuming a logarithmic velocity profile (Weber, 1999) : ( 2 21 ) Where is the mean hourly wind velocity, is the height above the ground, and is the friction velocity calculated per Weber (1999) : ( 2 22 ) where is the eddy covariance between the longitudinal and vertical fluctuating components. Surface roughness in fluences the boundary layer flow by decreasing the mean wind speed and increasing the turbulence intensity as the elevation above the ground decreases (Counihan, 1975) The change in mean wind speed and turbulence will affect t he temporal and spatial deposition of rain on the building faade (Blocken and Carmeliet, 2002) Building characteristics Immersion of a building in a wind flow creates turbulence in the form of frontal vortices, separations at the building edges/corners, corner streams, recirculation zones, shear layers and the far wake (Bottema, 1993) The flow pattern is dependent on

PAGE 38

38 upstream conditions, building orientation in the flow field, and building geometr ic shape. When raindrops approach the bluff body the trajectories become complicated; the result is non uniform deposition on the building facade. Trajectories of small particles change sharply; however, the higher inertia larger drops are less susceptible to local flow disturbances and bluff body aerodynamic effects. As a result, the deposition contours on the building faade of large drops are less affected than those corresponding to smaller drops ( Figure 2 6 af ter Blocken and Carmeliet, 2002). Figure 2 6 Deposition of smaller drop s left and larger drop s right (arrows indicate decreasing drop diameter) The effect of varying building geometry, in particular wid th to height ratios, changes the blockage effect on the wind flow. The number of drops diverted away from the structure increases with higher aspect ratios. Choi (1994) experimentally verified this phenomenon in an in vestigation of a narrow (H:W:D=4:1:1) building that exhibited higher LIF values than a wider (H:W:D= 4:8:2) building (assuming similar drop sizes). Modeling of Rain Deposition on the Building Faade Semi Empirical Models Measurements from vectopluviometers and surface mounted instruments have shown that is directly related to wind speed and horizontal rainfall intensity (Lacy, 1951; Hoppestad, 1955) This relationship has been the basis for multiple semi

PAGE 39

39 e mpirical models and has been used to derive regional WDR exposure from current standard meteorological data provided by existing weather stations. Hoppestad (1955) expressed the WDR intensity ( ) as a function of a WDR coefficient ( the invers e of drop terminal velocity), the wind velocity ( ), and the horizontal rain intensity ( ). His work provided the basis for current semi empirical models. ( 2 23 ) Lacy (1965) used empirical relationships that express the median raindrop size as a function of horizontal rainfall intensity and terminal velocity data to develop a single WDR coef ficient. The outcome was a refined equation that satisfies most WDR scenarios (Lacey, 1965) : ( 2 24 ) Empirical models have evolved to include the effect of the complex flow around the building and output spatially varying rain deposition rates. Two of t he most common models are the Straube and Burnett (2000) model and the British Standard BS EN ISO 15927 3 (2009) Both models implemen t : ( 2 25 ) where is the WDR coeffici ent dependent on location on building and is the angle between the wind direction and the perpendicular of the wall surface. A comprehensive comparison of these methods performed by Blocken et al. (2010) found that while the ISO model is more accurate t han the Straube and Burnett model, neither accurately model the blockage effect of the tested buildings.

PAGE 40

40 For the purpose of this study, wetting rates were estimated in accordance with the ISO model (BS EN ISO 15927 3:2009 ) : ( 2 26 ) Where is the rain intensity reaching the building faade, is the roughness coefficient representative of roughness of the terrain upwind of a wal l (conservatively assumed to be one), is the topography coefficient that accounts for wind speed up over isolated hills and escarpments (conservatively assumed to be one), is the obstruction coefficient that accounts for obstructions (e.g. build ings, fences, trees, etc.) close to and upwind of building faade (conservatively assumed to be one), and is the wall factor which is calculated as the ratio of water reaching the building faade to the quantity passing through an equivalent unobstructe d space ( conservatively assumed to be four, according to Table 4 in the BS EN ISO 15927 3:2009). Numerical Models The work by Choi (1993; 1994) has remained the foundation for most modern techniques. C hoi (1993, 1994) proposed a method for calculating WDR deposition on the building faade that includes: (1) calculating wind flow under steady state conditions by solving the Navier Stokes equations with the k calculating drop trajectories at every point for each raindrop size by iteratively solving their equations of motion 2 2 7 through 2 2 9 and 2 19 through 2 20 calculating and values at different locations of the building. ( 2 27 )

PAGE 41

41 ( 2 28 ) ( 2 29 ) In the equations of motion ( 2 27 2 2 9 ) is mass of the drop, is radius, is the air density, is the water density, is the air viscosity, and are along, across, and vertical wind directions, respectively. Blocken and Carmeliet (2002) extended this work by introducing a temporal component to the wind flow and examining the spatial and temporal WDR deposition on the facade of the VLIET ( Flemish Impulse Programme for Energy Technology ) test building (Blocken and Carmeliet, 2002) Full Scale Experiments Experimental wind engineering research directed at WDR effects on buildings is limited. Only a few projects have been conducted in the wind tunnel (Flower and Lawson, 1972; Rayment and Hilton, 1977; Inculet and Surry, 1994) and at full scale (e.g., Salzano, 2009; Bitsumala k, 2009 ; Lopez et al., 2011 ) Wind tunnel experiments have been used to characterize wetting patterns on the faade of multiple scale models Due their complexity and high cost, wind tunnel tests of WDR have been limited. Major difficulties encountered du ring these experiments include (1) short duration of tests to not saturate the water sensitive paper, (2) counting and measurement of drops on the water sensitive paper was extremely labor intensive, (3) large variability between tests as a result of short test durations, and (4) obtaining a uniform rain distribution was difficult due to the small size of the scaled drops (Inculet and Surry, 1994)

PAGE 42

42 Currently there are few methods for full scale simulation of WDR. These methods are primarily employed in the product approval process, at the state level, to ensure a minimum water infiltration resistance of building components. The testing procedures require the application of static and cyclic pressures while a constant wetting ra te is applied to the exterior of a singular building component; these tests do not directly investigate the holistic behavior of the building envelope, rather the behavior of each component in isolation. In response, research efforts by RDH Building Engine ering Limited (2002) Florida International University (Bitsuamlak et al., 2009) and the University of Florida (Masters et al.,2008; Salzano e t al., 2010; Lopez et al., 2011) to simulate WDR on full scale structures have been undertaken. RDH Building and Engineering Ltd. (2002) sought to identify the adequacy of building codes, standards, testing protocol s, and certification processes towards wind driven rain resistance of fenestration. The study analyzed the performance of 113 laboratory and 127 field window specimens. Field specimens were subjected to a constant impinging jet while a constant rain rate w as applied. The research at the University of Florida (Salzano et al., 2010; Lopez et al., 2011) expanded on the RDH study by evaluating water penetration resistance of window/wall assemblies subjected to wind loading calibrated to tropical cyclone field data collected by the Florida Coastal Monitoring Program (FCMP). This emerging sub discipline will greatly benefit from new insights regarding the physical simulation of WDR and can ultimately lead to improved performan ce evaluation of the water penetration resistance of building products and systems.

PAGE 43

43 Measurement of Wind Driven Rain Instrumentation In the earliest studies, WDR measurements were made using directional pluviometers (known as vectopluviometers) and wall mou nted collection chambers. Vectopluviometers are fixed instruments that obtain directional quantities of WDR in the free stream with four or eight compartments of similar size openings each facing the cardinal or cardinal and ordinal directions, respectivel y and an additional compartment facing the vertical direction (Lacy, 1951; Choi,1996) The implementation of vectopluviometers in various research projects (cf. Blocken and Carmeliet, 2004) has produced directional WD R data that has been used for the development of WDR maps (Lacy,1965) as a tool for estimating WDR deposition on building faades, and as basis for the assumption that the increases proportionally with wind speed and (discussed in semi empirical simulation methods). In the obstructed flow, collection chambers are attached to the walls of buildings to directly measures the quantity of WDR deposited on the f aade. These instruments are collection chambers, of multiple sizes, mounted flush to a wall with an attached reservoir to collect the deposited rain (Straube et al., 1995) Although implementation of these instruments facilita ted research of the spatial variability of WDR on the building surfaces and the effects of building aspect ratio, wind velocity, and horizontal rainfall intensity, they are unable to measure the RSD and wind characteristics. Modern precipitation measuremen t instruments more accurately determine the characteristics of rain but have limitations. Measurement by radar can produce detailed precipitation data for an extensive area from a single location; however, measurements are taken at heights where the RSD ca n vary from the ground level RSD, and the RSD

PAGE 44

44 is not directly measured. RSD determination by radar is based on reflectivity relationships (Richter and Hagen, 1997; Schafer et al., 2002) for which discrepancies have b een investigated by Marshall et al. (1947) Wilson and Brandes (1979) and Medlin et al. (2007) Disdrometers can accurately measure ground l evel RSD data at temporal resolutions that are useful for WDR research (Joss and Waldvogel, 1967; Loffler Mang and Joss, 2000) Impact and optical disdrometers are the two most commonly used rain sensors. Impact disdro meters, such as the Joss and Waldvogel disdrometer, measure the induced voltage from the displacement of an aluminum covered styrofoam sensor (Joss and Waldvogel, 1967) The voltage is amplified, and the drop size is interprete d by fitting the voltage to predetermined voltage ranges corresponding to drop diameters (Sheppard, 1990) The Joss and Waldvogel disdrometer can measure drop sizes between 0.3 mm and 5 mm with resolutions varying from 0.1 mm f or small drops to 0.5 mm for larger drops. Inherent limitations of the instrument manifest in high wind scenarios, because the measuring algorithm expects drops to be falling at terminal velocity (Sheppard, 1990 ; Tokay et al., 2008; Tokay et al.,2003) Optical disdrometers function by creating a light band and measuring the voltage drop from a photodiode, or series of photodiodes, as a drop passes through the light band and a fraction of the light is obstructe d from the photodiode (Loffler Mang and Joss, 2000) These instruments are capable of higher temporal resolution measurements due to the lack of a sensor head which requires time to return to its original position. Optical disd rometers are also more accurate than impact disdrometers in measuring particles less than 0.7mm (Loffler Mang and Joss, 2000)

PAGE 45

45 Limitations of optical d isdrometers Currently, optical disdrometers are the instrument of choice fo r measuring ground level RSD; however, these instruments have limitations that must be considered. The major limitations of optical disdrometers include fringe effects, coincident measurement of multiple drops, splashing effects, and errors associated with high wind. grazes the sensitive area; thus, the drop is recorded as a having an incorrect smaller diameter and travelling much faster than correctly measured drops. The erro r in velocity measurement propagates to an incorrect sample volume calculation which leads to an under estimation of and among other precipitation characteristics (Grossklaus et al., 1998) Errors associated with the coincident crossing of multiple drops occur if the optical disdrometer cannot discriminate between the light extinction from one or multiple drops. When multiple small drops simultaneously cross the sensitive area, a singular large drop is recorded travel ling at the velocity of the smaller drops. Thus, the sample volume is erroneously small and the incorrectly measured large drop has an exaggerated contribution to the rain parameters including and (Grossklaus et al., 1998) Splashing effects occur when drops strike the instrument, breakup, and the smaller remnants that bounce off, travel through the sensitive area at a much slower velocity than the termina l velocity (Tokay et al., 2001) The measured drops are not characteristic of the precipitation event and thus contaminate the data. The slow velocity results in a small sample volume and the splashed drops have an exaggerated contribution to the rain parameters including and

PAGE 46

46 The errors associated with high wind manifest as the incorrect diameter and velocity measurement due to the oblique trajectory of drops as they travel through the sample area. As wind velocities increase, the trajectory of the drops become s more horizontal; thus, they remain in the sensitive area longer and, through their passage, occupy more of the horizontally aligned light band, leading to larger sampled areas (Loffler Mang and Joss, 2000) Additionally, depe nding on the structure of the disdrometer, the distortion of the wind around the sensitive area significantly affects the trajectory of drops reducing the quantity of drops crossing the sensitive area (Nespor et al., 2000) Ma ny attempts have been made to mitigate the se effects (Donnadieu, 1980; Hauser et al., 1984; Grossklaus et al., 1998; Nespor et al., 2000; Tokay et al., 2001) Disdrometers have been develope d with the capability of sensing drops at the edge of the sampling area. Illingworth and Stevens (1987) and Grossklaus et al. (1998) developed disdrometers that employed an annular sensitive area rather than a flat sheet. With this advancement, drops that graze the sensitive area are characterized by a single voltage reduction and are thus excluded from the data. Drop Measurement Technologies implemented multiple photodiodes to meas ure the voltage drop across flat light sheet in the design of their Precipitation Imaging Probe (instrument is discussed in the next section). In this arrangement, when the outer most photodiodes sense a drop in the sensitive area, the drop is excluded fro m data. Other researchers (Donnadieu, 1980; Hauser et al., 1984; Tokay et al., 2001) mitigate the effect of fringe effects, coincident measurement of multiple drops and splashing effects limitations by emplo ying quality control algorithms that filter data according to the measured drop velocities and whether

PAGE 47

47 or not they fall within a velocity threshold based on the work done by Gunn and Kinzer (1949) These quality contro l algorithms typically exclude less than 20% of the data (Tokay et al., 2001) and yield realistic results. Errors associated with high wind speed, are addressed with the same algorithms and the exclusion of the data above a cer tain velocity threshold (Tokay et al., 2008) ; however, instruments employing an annular sensitive area, such as the Illingworth and Stevens (1987) and the Institut fur Meereskunde (IfM) dis drometer (Grossklaus et al., 1998) were designed for operation in high wind, but are not commercially available and to the authors knowledge, have only been employed on ships for rainfall measurement over open ocean. The instr ument platforms used in this research (described in Chapter 3), employed an OTT PARSIVEL disdrometer (Loffler Mang and Joss, 2000) and a Drop Measurement Technologies Precipitation Imaging Probe capable of high resolution measu rements, reporting precipitation at ten and one second intervals, respectively. The instruments are described in the following section. Instrumentation Used In This Study OTT PARSIVEL o ptical d isdrometer Loffler Mang and Joss (2000) produced the PARSIVEL an easy to handle, robust, and low cost disdrometer that lends the ability to implement a network of disdrometers to investigate small scale variability (Loffler Mang and Joss, 2000) The OTT PARSIVEL records the count, diamet er, and velocity of hydrometeors that pass through a 30 mm X 160mm X 1 mm laser field. It functions by focusing the laser on a single photodiode at the receiving end and measuring the analog voltage output ( Figure 2 7 ). As particles pass through the laser field they obstruct a band of light, corresponding to their diameter, from arriving to the photodiode and lower the measured

PAGE 48

48 voltage from the steady state 5 V. The voltage time history is inverted, amplified, filte red, and the DC component is removed (Loffler Mang and Joss, 2000) such that the r band, is used to estimate the particle velocity. Particle diameters and velocities are then binned into diameter and velocity classes ( Table 2 1 and Table 2 2 ), and digit ally output as time sampled matrices ( Figure 2 8 ). The particle shapes are assumed to be symmetric in the horizontal plane and have linearly varying axis ratios of 1 to 1.3 for diameters 1 mm to 5 mm, 1 (spherical) for diameters < 1 mm and 1.3 for diameters > 5 mm. Table 2 1 PARSIVEL diameter classes Class Number Class Average (mm) Class Spread (mm) 1 0.062 0.125 2 0.187 0.125 3 0.312 0.125 4 0.437 0.125 5 0.562 0.125 6 0.687 0.125 7 0.812 0.125 8 0.937 0.125 9 1.062 0.125 10 1.187 0.125 11 1.375 0.250 12 1.625 0.250 13 1.875 0.250 14 2.125 0.250 15 2.375 0.250 16 2.750 0.500 17 3.250 0.500 18 3.750 0.500 19 4.250 0.500 20 4.750 0.500 21 5.500 1.0 00 22 6.500 1.000 23 7.500 1.000 24 8.500 1.000

PAGE 49

49 Table 2 1 Continued Class Number Class Average (mm) Class Spread (mm) 26 11.000 2.000 27 13.000 2.000 28 15.000 2.000 29 17.000 2.000 30 19.000 2.000 31 21.500 3.000 32 24.500 3.000 Table 2 2 PARSIVEL velocity c lasses Class Number Class Average (m/s) Class Spread (m/s) 1 0.050 0.100 2 0.150 0.100 3 0.250 0.100 4 0.350 0.100 5 0.450 0.100 6 0.550 0.100 7 0.650 0.100 8 0 .750 0.100 9 0.850 0.100 10 0.950 0.100 11 1.100 0.200 12 1.300 0.200 13 1.500 0.200 14 1.700 0.200 15 1.900 0.200 16 2.200 0.400 17 2.600 0.400 18 3.000 0.400 19 3.400 0.400 20 3.800 0.400 21 4.400 0.800 22 5.200 0.800 23 6.000 0.800 24 6. 800 0.800 25 7.600 0.800 26 8.800 1.600 27 10.400 1.600 28 12.000 1.600 29 13.600 1.600 30 15.200 1.600 31 17.600 3.200 32 20.800 3.200

PAGE 50

50 Figure 2 7 PARSIVEL d rop diameter and velocity d etermination PARSIVEL Output: Class V Class D 1 2 3 4 5 6 7 32 1 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 3 0 0 5 0 0 0 0 0 0 4 0 0 0 0 0 0 0 0 0 5 0 0 0 0 0 0 0 0 0 6 0 0 0 0 0 0 4 0 0 7 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 32 0 0 0 0 0 0 0 0 0 Figure 2 8 PARSIVEL m easurement and o utput

PAGE 51

51 Droplet M easurement Technologies Precipitation Imaging Probe The Droplet Measurement Technologies (DMT) Precipitation Imaging Probe (PIP) is a 2 D optical dis drometer that functions in the same manner as the OTT PARSIVEL but has a higher resolution due to the use of 64 photodiodes instead of one. It measures and counts drop diameters with 0.1 mm resolution up to 6.2 mm in a sample area of 260.0 mm x 6.4 mm. The PIP does not measure drop velocity; it assumes that all drops are moving at the wind velocity and thus requires a continuous wind velocity input. Additionally, the PIP is intended to be mounted on aircraft and operated at flight speeds. Thus, the sample a rea is smaller than the OTT PARSIVEL sample area leading to a smaller sample volume. Rain parameters for both disdrometers are calculated as follows: ( 2 30 ) ( 2 31 ) ( 2 32 ) ( 2 33 ) ( 2 34 ) ( 2 35 )

PAGE 52

52 ( 2 36 ) The variables introduced in Equations 2 30 to 2 36 are defined as follows: is the number of diameter drops is the sample volume is the depth of field of the instrument is the width of the diode array is the velocity of drop through the laser plane is the sample time of the instrument is the density of water is the third moment as defined in ( 2 8 is the equivalent radar reflectivity is the mean measured drop diameter is the volume weighted mean drop diameter (Testud et al., 2001) Summary This chapter presented an overview of the WDR phenomenon, simulat ion methodologies adopted by other researchers and instruments used to measure precipitation Information of the disdrometers used in this research has been presented, and their applicability for WDR measurement will be discussed in Chapter 3.

PAGE 53

53 CHAPTER 3 DESIGN, PROTOTYPING, AND IMPLEMENTATION OF ARTICULATING RAIN PARTICLE SIZE MEASUREMENT PLATFORMS This chapter presents the development and implementation of novel instrument platforms intended for characterizing the raindrop size distribution (RSD) in strong winds wind induced errors associated with ground based disdrometers. Thus, the data collected provides a new resource for the study of wind driven rain (WDR). This chapter is orga nized into three sections. First, the motivation for the development of an instrument platform which actively aligns the disdrometer perpendicular to the mean rain direction is discussed. This section demonstrates that in the absence of wind (i.e., drops t ravelling perpendicular to the sensitive area) measurements made by the PARSIVEL disdrometer and measurements made using a laborious but highly accurate procedure, the Oil Medium Test, are similar. This section also demonstrates the limitations of the inst rument in the presence of wind (i.e., drops travelling obliquely through the sensitive area). Second, details of the WDR measurement system are presented. Finally, a comparison of collocated stationary and actively aligned disdrometers (henceforth referred to articulating instruments) is presented. Motivation for the Development of an Articulating Instrument Platform Optical disdrometers are the instrument of choice for quantifying RSDs; however, due to the limitations of these instruments (discussed in Ch apter 2), data recorded in high winds are usually excluded (Tokay et al., 2008) Lack of high quality data has left questions regarding the character of WDR in high wind events. Griffiths (1975) proposed that actively aligning a disdrometer parallel to the mean rain direction could theoretically improve accuracy by showing that the sample volume of an articulating

PAGE 54

54 sensor was greater than that of the sensor in the stationary, horizontally aligned, configuration (the proof is show n in the Appendix). Bradley and Stow (1975) attempted such an experiment but found that device was unmanageable in high wind, dripped water into the sensitive area when tilted, and were not certain how to define a mean rain direction. Since then, disdrometers and servomotor systems have become lighter, easier to use, and less expensive. Thus, the author investigated the option of aligning disdrometers to the mean rain direction by performing a series of laboratory tests intended to reduce measurement errors associated with high wind speeds. R aindrop S ize D istribution Verification Using the Oil Medium Test The Oil Medium Test was performed to demonstrate that when properly employed (i.e. drops travelling perpendicular to the sample area), the PARSIVEL disdrometer yields a reliable estimate of the RSD. The test was performed under ideal stagnant air conditions where the RSDs recorded by the PARSIVEL were compared to RSDs obtained from water drops collected in oil. The PIP w as not used to calculate the stagnant air RSD because the instrument requires continuous wind velocity feed back to calculate the sample volume, rather than using the drop velocity. Raindrop size distributions from drops collected in oil were calculated u sing a morphological image processing algorithm (van den Boomgaard and van Balen, 1992; Adams, 1993; Jones and Soille, 1996) The experimental setup included setting a single nozzle 3.0 m above a collection di sh containing the oil medium, setting the water pressures, and exposing the collection dish to five seconds of continuous spray. A 20 megapixel, RAW format, picture of the collection dish was then immediately taken using a Canon EOS 5D Mark II Digital SL R Camera with a Canon Telephoto EF 100mm f/2.8 USM macro lens with the ISO set to 320, relative aperture set to 7.1, white balance

PAGE 55

55 set to 6300K, and shutter speed set to 1/10 sec. To maximize the contrast between the dish and drops, a blue dye (Cole Par mer #00298 18) was used. Care was also taken in the selection of the oil medium, by selecting a clear mineral oil with a specific density similar to water. The oil properties prevented distortion of the shape of the drops and essentially captured the drops as they were when falling freely. The photographs were filtered using Adobe Photoshop to obtain a high contrast black and white image. The black and white images were then analyzed by the morphological image processing algorithm to obtain the size of ea ch drop by scaling the pixel counts ( Figure 3 1 ). The data was then binned using PIP specifications and an RSD was obtained. In this experiment, six measurements were taken by the PARSIVEL and images of 15 oil medi um tests were processed by the morphological image processing algorithm, yielding the RSDs shown in Figure 3 2 In general the PARSIVEL measured a smaller number of drops larger than 2 mm in diameter; however, the source of the difference is unclear and for practical purposes the two RSDs are reasonably similar. Figure 3 1 Sample picture from morphological image processing algorithm

PAGE 56

56 Figure 3 2 Comparison of different raindrop size distribution measurement techniques Bearing Drop Test The bearing drop test was performed to replicate the effect of high wind speeds on stationary PARSIVEL measurements by dropping 20 high tolerance steel bearings(1 mm, 2 mm, and 3 mm) from 6.1 m height ( Figure 3 3 ) through the laser plane at multiple elevation angles (90, 80, 70, 60, 50, 45, 40). The results of this test are found in Figure 3 4 and Figure 3 5 The figures demonstrate that the PARSIVEL correctly measured the bearing diameters regardless of trajectory angle; however, at 50 or lower, incorrect m easurements of the velocity indicate that the measured velocity is dependent of the trajectory angle and can deviate significantly from the theoretical velocity. Small diameter measurements (much less than 1mm) are assumed to be margin fallers. Similar res ults were observed by Friedrich et. al (2011) in which 2 6 mm drops were recorded as having reduced fall velocities when travelling obliquely through the sensitive area; however, at more extreme elevation angles, unrealistically high drop diameters were re corded indicating that incorrect drop diameter measurements may

PAGE 57

57 occur high wind speeds, even though it was not observed in this experiment. Thus, the observed results may not be representative of every natural scenario. Figure 3 3 Angled trajectory experiment configuration To relate the trajectory angle to specific wind speeds, 3 1 was used (Based on the Lacy 1965 Equation): ( 3 1 ) where is the terminal velocity of the drop and is the wind velocity. Table 3 1 demonstrates the sensitivity of the trajectory angle to drop diameter and wind velocity. The table indicates that at 10 m/s wind speeds, trajectory angles of the majority of the drops (< 5 mm) will be far below the 50 threshold. This implies that at wind speeds of 10 m/s drop velocities can be underestimated by at least 40%, yieldi ng incorrect RSD measurements. The effect of oblique trajectory angles was further investigated in winds generated by the UF Windstorm Simulator. Table 3 1 Trajectory a ngles of m ultiple d iameter d rops in m ult iple w ind s peeds Drop Diameter (mm) V t (m/s) U = 5 m/s U = 10 m/s U = 15 m/s U = 20 m/s 1.0 4.0 38.9 21.9 15.0 11.4 2.0 6.5 52.4 33.0 23.4 18.0 3.0 8.1 58.2 38.9 28.3 21.9 4.0 8.8 60.5 41.4 30.5 23.8 5.0 9.1 61.2 42.3 31.2 24.4

PAGE 58

58 Figure 3 4 PARSIVEL m easured diameters at multiple trajectory angles

PAGE 59

59 Figure 3 5 PARSIVEL m easured velocities at multiple trajectory angles (dashed line ind icates theoretical velocity)

PAGE 60

60 Instrument Performance in High Wind Speeds The option of orienting the PARSIVEL into the mean rain direction was further investigated by recording the RSD with PARSIVELs oriented at 0.0, 22.5, and 45.0 elevation angles; a no zzle array operating at 68.9, 103.4, and 137.9 kPa water pressures (corresponding to approximately 400, 475, and 550 mm/hr); and steady wind velocities of 8.9, 17.9, and 35.8 m/s. Three PARSIVEL measurements were taken for each test configuration listed in Table 3.2. The PIP measurements were used as the reference measurement, given that this instrument is intended to be used in high wind speeds and is more accurate. The instruments were mounted on a gantry system shown in Figure 3 6 The PARSIVEL disdrometer exhibited no noticeable inaccuracies due to increasing rainfall intensities; however, inaccuracies were observed at the high wind velocities ( Figure 3 7 ). As expected, a t high wind velocities, the instruments oriented at 22.5 and 45.0 recorded drops travelling slower than the instrument at 0.0 ( Figure 3 8 ). The incorrectly measured velocities decrease the sample volume over whi ch the RSDs are calculated; therefore, data from these instruments yield incorrect RSDs (as shown in Figure 3 7 ). This experiment also demonstrated that the PARSIVEL measured a smaller number of large drops (larger than 2 mm) when compared to the PIP; however, for practical purposes, when the PARSIVEL is properly employed, the measured RSD is similar to that of the PIP. These results demonstrate that, as Griffiths (1974) propose d, aligning the disdrometer with the mean rain direction improves accuracy. Thus, a novel approach was taken to continuously align the disdrometer with the mean rain vector; described in the following section.

PAGE 61

61 Table 3 2 Steady w ind t est m atrix Test Number Angle (deg) Wind Speed (m/s) Water Pressure (kPa) 0.0 22.5 45.0 8.9 17.9 35.8 69 103 138 1 2 3 4 5 6 7 8 9 Figure 3 6 Steady wind instrument configuration (p hoto courtesy of author )

PAGE 62

62 Figure 3 7 PARSIVEL m easured raindr op size distribution s for the tests in steady wind flow

PAGE 63

63 Figure 3 8 PARSIVEL m easured drop diameters and v elocities for the tests in steady wind f low

PAGE 64

64 Articulating Instrumentation Platforms An instrumentat ion platform was designed, prototyped, and implemented in supercell thunderstorms and Atlantic hurricanes to supplement FCMP weather station T 3 which contains a similar actively controlled system that aligns the PIP to the mean rain direction. The platfor ms are capable of high temporal resolution measurements of wind and rain. This section describes the components of the articulating instrumentation platform (illustrated in Figure 3 9 ), its operation, the quality c ontrol algorithm, and a description of FCMP weather station T 3. Articulating Instrumentation Platform Components The articulating instrument platform consists of the following six components, which are discussed in the following subsections: 1. An OTT PARSIV EL optical disdrometer (described in Chapter 2) 2. An RM Young Model 851062D Sonic Anemometer 3. An articulating instrument support structure that is driven by two IMS M 17 stepper motors with an integrated controller and encoder ( Model MDI4MRQ17C4 EQ G1C3 ) 4. A La bview 8.5 software data acquisition system, which consists of a laptop computer connected to a four port RS 485 National Instruments serial interface (Model No. NI USB 485/4 ) 5. A battery array of 12V 35 AH Power Sonic (PS 12350 NB) batteries to supply power to the instruments 6. A substructure to provide lateral stability and to transfer the gravity and wind loads to the ground

PAGE 65

65 Figure 3 9 Articulating instrument platform RM Young s onic a nemometer Wind velocity a nd direction are measured by an RM Young Model 85106Sonic Anemometer. A sonic anemometer was chosen for its compact size, lack of moving parts, and superior performance in low winds when compared to a mechanical (plate or cup) anemometer. Mechanical anemom eters have a distance constant on the order of 2 10m, and for practical purposes, the distance constant of sonic anemometers is considered to be zero. Sonic instruments operate by measuring the time it takes for ultrasonic pulses to travel between four tra nsducers. The 85106 is capable of measuring wind velocities up to 70 m/s 0.1 m/s and direction 0 to 360 2 at 4 Hz. Data are output on independent analog channels (0 5V) and on a digital channel to the motion control systems and computer, respectively

PAGE 66

66 Articulating support s tructure The instrumentation array is continuously rotated by two Intelligent Motion Systems (IMS M17Plus) motion control systems (stepper motor, driver and controller) to align the optical disdrometer with the ten second mean rain vector ( Figure 3 10 ). The motor controllers are programmed in a proprietary programming language (MCode) to read the 0 5V analog signal for wind speed and direction from the sonic anemometer. The voltage is interpr eted as the angle required to rotate the instrument. The elevation equations were based on the Lacy relationship (1970, Equation 2.6) for a 1.2 mm particle. Figure 3 10 Articulat ing instrument platform automation To ensure that the platform will resist high wind loads it was designed to withstand the expected peak gust wind load caused by a minimal Category 5 tropical cyclone ( ), as defined by the Saffir Simpson Scale (Simpson 1974, Simpson and Wind/Rain Direction Parsivel will slowly rotate into alignment with the mean rain vector Instrument array will slowly rotate into alignment with the mean wind velocity vector

PAGE 67

67 Riehl 1984). The structural system was idealized as a system of cylinders acted on by drag forces computed using : ( 3 2 ) ( 3 3 ) Where is air density, is air velocity, is cylinder diameter, is cylinder length, is drag coefficient, and is kinematic viscosity. Drag coefficients were found using Figure 3 11 and ( 3 3 for a given Reynolds number, The moment at w as found to be while the moment required to overturn the instrument was found to be Figure 3 11 Drag coeffici ent of a cylinder dependant of R eynolds number of flow (Adapted from R Pant on, Incompressible Flow 3rd ed.) Data a cquisition s ystem Data from the disdrometer, anemometer, and internal encoders on the stepper motors are transmitted digitally via RS 485 lines and recorded on a laptop computer running a custom written data logger op erating on the National Instruments LabView 8.5 platform. The control and data diagram is shown in Figure 3 12 The system is also

PAGE 68

68 capable of streaming summary data via a cellular connection, should future needs wa rrant an upgrade. Figure 3 12 Inst rument platform c ontrol and data d iagram Power The system was designed to run on battery power for a period of time sufficient to record the passage of a tropical cyclone. With capacity of each, three Power Sonic (PS 12350 NB) batteries will supply the mean amperage draw of 4 amps for over 24 hrs. Substructure The instrument platform is placed above a heavy duty tripod with wide set legs to maximize the overturn capacity. At the base of each tripod leg a 609.6 mm (24 in) steel stake is embedded in the ground. A guide wire can also be used to secure the system to an earth screw placed below the center of gravity. Once the tripod is secured, the

PAGE 69

69 data and power cables are connected t o ports on the enclosure box and the software is set to operate. portability and rapid deployment. During the VORTEX2 deployment the instrument was assembled, operational, a nd recording in less than three minutes. Portability of the instrument array is required to allow the set up of multiple systems in storm paths expected to contain the most rain. Rapid deployment is crucial since weather conditions can quickly worsen and p ose unsafe conditions for the operator. Portable Instrument Platform Operation System operation, as illustrated in Figure 3 12 begins with anemometer measurement of wind velocity and direction. The anemometer outp uts the 0 5 V analog signals to the motion control systems which continuously calculate the 10 second average. The azimuth motion control system interprets the deviation from 2.5V the voltage corresponding to a wind direction perpendicular to the instrum entation platform as a number of steps for the stepper motors to rotate. When the motion is complete the control system re samples and rotates. This process continuously aligns the instrument array perpendicular to the wind direction. The elevation motion control system interprets the analog signal and assigns a number of steps to rotate based on the programmed formula relating wind velocity to drop trajectory (Lacy 1970, Equation 2 6 for 1.2 mm particle). This aligns the disdrometer to the mean rain vecto r. The motion control systems digitally output their location, via RS 485 channels, at the completion of each prescribed rotation. The disdrometer and anemometer digitally output their measurements via RS 485 channels every 10 seconds and at 14 Hz, respect ively. The

PAGE 70

70 data acquisition system records the measurements from all of the instruments, as previously described. Quality Control Algorithm To mitigate fringe effects, coincident measurement of multiple drops and splashing effects (described in Chapter 2) a quality control algorithm was employed. The algorithm filters data that fall below 50% of the expected terminal velocity (Gunn and Kinzer, 1949) and above the maximum expected velocity ( ), defined as the resultant of the terminal velocity ( ) and wind speed ( ) plus one standard deviation Additionally, the algorithm only considers data above 5 mm/hr and drops smaller than 10 mm ( Figure 3 13 ). The algorithm ensures that drops that graze the sensitive area (characterized by small measured diameter at a high speed), multiple drops (characterized by a large drop at low speed), and drops resulting from splashing (characterized by small dro ps at low speed) are mostly excluded from the analyzed dataset. A) Observed

PAGE 71

71 B) Figure 3 13 Quality control filter at U = 0 m/s and U = 10 m/s Description of Weather Station T 3 The FCMP portable weather stations are 10 m structural steel lattice towers mounted on dual axel trailers for rapid deployment (Poss, 2000) The tower system and outriggers unfold and are operational in minutes. The stations are equipped with three fixed axis a nemometers (RM Young 27106R) at 5 m and 10 m to measure the 3D wind speed and direction. At the 10 m location is an additional anemometer (RM Young 05103V) that measures the resultant of the lateral and longitudinal component of wind and serves as a redund ant measurement (Balderrama et al., 2011) FCMP tower T 3 is the only of the six weather stations that is outfitted for precipitation measurement. Rainfall measurements are made by a DMT PIP (described in Chapter 2) at a 3 m he ight. In order for the PIP operate correctly, the laser plane must be oriented perpendicular to the rain vector. To achieve this, the PIP is mounted on an automated Observed

PAGE 72

72 turret that continuously aligns the disdrometer ( Figure 3 14 ). The azimuth angle of the turret is actively controlled by a positioning system that continuously samples the gill anemometers at 5.0 m. The elevation angle is governed by the elevation equations derived from work by Lacy (1970, Equation 2.6 f or 1.2 mm particle). Wind and rain are reported at 10 Hz and 1Hz, respectively. Results from data gathered by both platforms during supercell thunderstorms and Atlantic hurricanes are discussed at length in the Chapter 4. Figure 3 14 T 3 P recipitation Imaging P robe t urret system and Gill anemo meters at 5.0 m (p hoto courtesy of author )

PAGE 73

73 Comparison of Collocated Stationary and Articulating Instruments in a Supercell Thunderstorm Data from a supercell thunder storm were collected with collocated stationary and articulating instrumentation platforms (Figure 1 1). A difference in accuracy was confirmed in measurements of the RSDs between the two instrumentation platforms. The stationary instrumentation platform m easured unrealistic large number concentrations of drop diameters larger than 4 mm ( Figure 3 15 top) in high wind velocities (> 10 m/s). This phenomenon was not apparent in the articulating instrument platforms ( Figure 3 15 center). The unrealistic large number concentrations were found to be attributable to erroneously low measured drop velocities (generally less than 2 m/s). Incorrect measurements of the RSD subsequently l ead to incorrect estimates of all rainfall parameters in the stationary instrumentation platforms (rainfall intensity Figure 3 16 and reflectivity Figure 3 17 ). The art iculating instrument platform did not exhibit similar jumps in the data, indicating reliable measurements. A)

PAGE 74

74 B) C) Figure 3 15 Measured raindrop size distribution by stationary instrumentations ( A ) and a rticulating instrumentation ( B ) Figure 3 16 Comparison of rainfall intensities measured by stationary and articulating instrument platforms

PAGE 75

75 Figure 3 17 Co mparison of estimated reflectivity by stationary and articulating instrument platforms Summary This chapter presented the known inaccuracies in RSD characterization c a used by strong w inds As a result, an articulating instrument platform was designed, prot otyped, and evaluated in the field. The articulating instrument platform did not exhibit the errors observed in the stationary instrument; thus, the approach taken to measure WDR in strong winds was validated. The next chapter presents the analysis and res ults of data collected during the VORTEX2 and FCMP field campaigns.

PAGE 76

76 CHAPTER 4 CHARACTERIZATION OF WIND DRIVEN RAIN IN STRONG WINDS This chapter addresses field research conducted during the Verification of the Origins of Rotation in Tornadoes Experiment 2 (VORTE X2) and Florida Coastal Monitoring Program (FCMP) field deployments during Hurricanes Ike (2009) and Irene (2011). This portion of the research analyzed the wind driven rain (WDR) data collected in extreme events. The goals were to: 1. Determine the effect of wind on raindrop size distribution (RSD) 2. Determine if the RSD models used in computational wind engineering are appropriate for modeling the rain deposition rate on buildings during a design level wind event 3. Characterize the peak to mean ratio of rainfall intensity and determine the impact on design for water penetration resistance 4. Investigate if the precipitation algorithm in the radar product generator used by the National Weather Service (NWS) Weather Surveillance Radar exhibits any biases that manifest in extreme wind events This chapter is organized into s ix sections: (1) a description of the field research programs, (2) the effect of wind on the drop diameter, (3) the effect of wind on the RSD, (4) comparison of RSD models to RSDs measured in multipl e wind velocities, (5) calculation of the peak to mean ratios of rainfall intensity ( ), and (6) comparison of mea sured rain data to WSR 88 D data Field Research Programs Verification of the Origins of Rotation in Tornadoes Experiment 2 (VORTEX2) Overview The Verification of the Origins of Rotation in Tornadoes Experiment 2 (VORTEX2) Project a continuation of the original VORTEX project described by Rasmussen et al. (1994) was an interdisciplinary multi agency effort t o investigate tornado genesis,

PAGE 77

77 dynamics, kinematics, demise, supercell near ground wind field, and how the environment regulates storm structure. VORTEX2 assets included 10 mobile radars, 12 mobile mesonet instrumented vehicles, and 38 deployable instrumen ts including, disdrometers, surface level wind measurement stations, weather balloon launching vans, and unmanned aircraft that were deployed in the projected path of supercell thunderstorms minutes prior to their arrival. Data collected will assist scient ists in better understanding tornadic behavior and improving tornado forecasting. Deployment d etails As part of VORTEX2, a disdrometer team was tasked with the repeated deployment of eight instrument stations (articulating and stationary) for the collecti on of RSD data. During each day of field collection, team leaders would determine locations exhibiting favorable conditions for storm formation. The teams would then mobilize and target a supercell thunderstorm that exhibited potential for tornadic activit y. The instruments were deployed perpendicular to storm motion in the path of the hook appendage minutes before the approaching storm passed through the area ( Figure 4 1 ). Spacing between instrument stations was 0. 5 km to 2 km based on storm velocity and the projected storm path. Upon storm passage instruments were collected and relocated in the projected path of the same storm in the same configuration. A total of 144 measurements were collected during 32 supercell thunderstorms with eight instrument platforms ( Table 4 1 ) in the states shown in Figure 4 2 through the period of May 1 to June 15, 2010. Data admitted for analysis consis ts of the 25 observations collected with the articulating instrumentation platforms ( denoted by in Table 4 1 ); the observations consist of 16 Hz wind velocity and 0.1 Hz rain time histories (see Chapter

PAGE 78

78 3). The d ata was collected over open terrain, clear of any obstructions, and at a sufficient distance from roads to minimize errors associated with vehicle spray. Table 4 1 Verification of the rigins of Rotation in T ornadoes Experiment 2 d eployment d etails Date Time (UTC, HHMM) Location Instrument Platforms 06 May 0010 0058 Oberlin, KS CU1 CU2 10 May 2212 2303 Chandler, OK CU1 CU2 11 May 0128 0240 Clinton, OK CU1 CU2 UF1 UF4 UF5 UF6 UF7 12 May 2350 0100 Willow OK CU1 UF4 UF5 UF6 UF7 14 May 1850 1900 Odessa, TX UF7 14 May 1934 2030 Odessa, TX CU1 UF4 UF5 UF6 UF7 14 May 2127 2230 Odessa, TX CU1 UF4 UF5 UF6 UF7 15 May 2330 0800 Artesia, NM CU1 UF4 UF5 UF6 UF7 17 May 2150 2345 Artesia, NM CU1 UF1 UF4 UF5 UF 6 UF7 18 May 2240 0300 Dumas, TX CU1 UF1 UF4 UF5 UF6 UF7 19 May 2029 2145 Watonga, OK CU1 21 May 0035 0123 Mitchell, NE CU1 UF1 UF4 UF5 UF6 UF7 24 May 2123 2205 Ogallala, NE CU1 UF1 UF4 24 May 2230 2300 Ogallala, NE CU1 UF4 UF5 UF6 2 5 May 0005 0100 Ogallala, NE CU1 UF5 UF6 UF7 29 May 0015 0140 Thedford, NE CU1 UF1 UF3 UF4 UF5 UF6 UF7 0 2 Jun 2255 2355 Benkelman, NE CU1 UF1 UF3 UF5 UF6 0 3 Jun 0020 0100 Wagner, SD UF1 UF3 UF5 UF6 0 5 Jun 2330 0010 Des Moines, IA CU1 UF1 UF3 UF5 UF6 0 6 Jun 2220 0210 Ogallala, NE CU1 UF1 UF5 UF6 0 6 Jun 0110 0210 Ogallala, NE CU1 UF1 UF5 UF6 0 7 Jun 2335 0030 Scottsbluff, NE CU1 UF1 UF3 UF5 UF6 UF7 08 Jun 0100 0200 Scottsbluff, NE CU1 UF5 UF6 UF7 0 9 Jun 0040 0240 Scottsbluff, NE CU1 UF1 UF3 UF5 UF6 UF7 10 Jun 2330 0045 Wiggins, CO CU1 UF1 UF3 UF5 UF6 UF7 11 Jun 2306 0030 Limon, CO CU1 UF1 UF3 UF5 UF6 UF7 12 Jun 2050 2300 Gruver, TX CU1 UF1 UF5 UF6 UF7 13 Jun 1830 1914 Darrouzett, TX CU1 UF1 UF6 13 Jun 2050 2140 Darrouzett, TX CU1 U F5 UF6 UF7 13 Jun 2250 2340 Darrouzett, TX CU1 UF5 UF6 UF7 *14 Jun 2000 2042 Post, TX CU1 UF5 UF6 UF7

PAGE 79

79 Figure 4 1 VORTEX2 i nstrument dep loyment Figure 4 2 VORTEX2 d ata collection sites Florida Coastal Monitoring Program (FCMP) Overview The Florida Coastal Monitoring Program is a collaborative research program between multiple universities (University of Florida, Clemson University, Florida International University, and Florida Institute of Technology) and the insurance industry (IBHS) focusing on the full scale experimental study of tropical cyclone ground level

PAGE 80

80 wind fields and wind loads on residential structures (Balderrama et al., 2011) Since 1998, the FCMP has deployed portable weather stations to collect ground level meteorological observations and roof pressure sensors on single family residences to measure wind induced roof uplift pressures in tropical cyclones. FCMP da ta are used widely by meteorologists and emergency management, as well as university researchers developing CFD and numerical models, and conducting boundary layer wind tunnel and experimental full scale experiments. Deployment d etails During the 2008 and 2011 Atlantic hurricane seasons, the FCMP deployed WDR measurement stations for Hurricanes Ike (2008) and Irene (2011). Three data records were collected in the Greater Houston Area, the Outer Banks of North Carolina, and Deal, New Jersey. The following is a brief narrative of the deployments. Ike became a tropical storm on September 1, 2008, approximately 1300 km west of the Cape Verde Islands and steadily intensified to a Category 4 storm over the following two days as it moved west northwestward over th e tropical Atlantic. By September 7 Hurricane Ike made landfall at the southeastern Bahamas and by September 8 its center had reached Cabo Lucrecia, Cuba. The storm emerged from Cuba near Cabo Lucrecia into the Gulf of Mexico on September 9th as a Category 1 storm. Hurricane Ike slowly intensified through the gulf and made landfall at the north end of Galveston Island, Texas as a Category 2 storm on September 13. FCMP assets (including weather station T 3) were deployed in Baytown, Texas (29. 801944 N, 94. 88221 W) approximately seven hours prior to landfall. The weather station recorded 18 hours of wind and rain data prior to, during, and after eye wall passage. The terrain

PAGE 81

81 exposure for the weather station was primarily open to the east and suburban elsewhe re. Topography for the region was flat. Irene became a tropical storm on August 20, 2011, approximately 440 km East of Martinique. As it travelled West Northwest Irene intensified into a Category 1 hurricane on August 22, 50 km north of Isabela, Puerto Ri co. By August 24, the storm quickly intensified into a Category 3 hurricane as it passed over the Turks and Caicos Islands and the Bahamas. By August 26, Irene had passed over the Bahamas and was approximately 300 km east of Florida. As Irene continued nor thward it encountered unfavorable conditions and weakened to Category 1 storm before making landfall less than 5 km from Beaufort, NC at approximately 1200 UTC on August 27. The FCMP deployed an articulating instrument platform in Beufort, NC (34. 720525 N 76. 639568 W) approximately eight hours before landfall. The articulating instrument platform recorded 15 hours of wind and rain data prior to, during, and after eye wall passage. While the articulating instrument platform recorded data, a second team co ntinued northward towards Deal, NJ (40. 253326 N, 73. 991427 W) where weather station T 3 was deployed at 0300 UTC on August 28. Weather station T 3 recorded 13 hours of wind and rain data approximately 11 km from the storm center as it weakened to tropica l storm status. The terrain exposure for the articulating instrument platform was primarily suburban in all directions. Terrain exposure for weather station T 3 was primarily open to the east and suburban elsewhere. Topography for both regions was flat. T o compare data collected by weather station T 3 to data gathered using the articulating instrument platform, 10 second segmental averages of disdrometer data and

PAGE 82

82 the mean velocity ( ), longitudinal turbulence intensity ( ), and lateral turbulence intensity ( ) for the 60 seconds leading up to each segment were calculated. The following section investigates the effect of wind on raindrop diameter. Effect of Wind Velocity and Turbulence Intensity on Raindrop Diameter The effects of longitudi nal wind velocity ( ), longitudinal turbulence intensity ( ) and lateral turbulence intensity ( ) on raindrop sizes were investigated by comparing the shapes of calculated probability density functions (PDFs) of raindrop diameter ( ), me an raindrop diameter ( Equation 2 35 ), and volume weighted mean raindrop diameter ( ,Equation 2 36 ). Turbulence intensity is a measure of the variation in wind speed about the mean and is defined as the coefficient of variation (i.e., ratio o f the standard deviation and mean) : ( 4 1 ) ( 4 2 ) The standard deviations of the longitudinal and lateral wind components are respectively denoted as and and the mean longitudinal wind speed is denoted as It was hypothesized that rate of break up and coalescence may be affected by turbulence; subsequently, the drop diameters would be rel ated to turbulence intensity. Typically turbulence intensities are calculated using one hour records (Harper et al, 2009); however, a rain event can fluctuate significantly over a much shorter time period; thus, 60 sec segments were selected for this analy sis. Calculated PDFs of VORTEX2 data indicate that and ( Fig ure 4 3 ) are dependent on The series of and PDFs show that as the longitudinal wind speed increases, standard deviation decreases, and the mode shifts left. This behavior

PAGE 83

83 in dicates that probability of smaller and increases as the longitudinal wind speed increases. The relationship is further evident in observing the 0.25, 0.50, and 0.75 quantiles of and which demonstrate that there is a general n egative trend of and as the longitudinal wind speed increases. PDFs and quantiles of VORTEX2 data also indicate that as both and increase, increases slightly. Similar trends were observed in the FCMP dataset. The PD Fs of are shown in Figure 4 6 ; the probability of a smaller increases as the longitudinal wind speed increases. Similarly, the quantile plots indicate a positive trend between and longitudinal wind s peed. Turbulence intensity trends were also observed in this dataset. and were observed to decrease in the presence of higher and ( Figure 4 7 and Figure 4 8 ). The remaining figures of both datasets indicate that does not significantly change with or The increase in or with increased or the decrease in or in the presence of highe r or is attributed to one of the following issues: (1) it is possible that in increased wind velocities the high drag forces acting on the drops overcome the surface tension holding the shape together and breakup occurs or (2) high turb ulence intensities (either or ) cause the probability of coalescence to increase. Given that these trends were observed, the next step in the research was to investigate the effect of wind on the RSD, discussed in the next section.

PAGE 84

84 Fig ure 4 3 Effect of longitudinal wind velocity on drop diameter observed in VORTEX2 dat a

PAGE 85

85 Figure 4 4 Effect of longitudinal turbulence intensity on drop diamet er observed in VORTEX2 data

PAGE 86

86 Figure 4 5 Effect of lateral turbulence i ntensity on d rop d iameter o bserved in VORTEX2 d ata

PAGE 87

87 Figure 4 6 Effect of longitudinal wind velocity on drop diameter observed in FCMP d ata

PAGE 88

88 Figure 4 7 Effect of longitudinal turbulence intensity on drop diameter observed in FCMP da ta

PAGE 89

89 Figure 4 8 Effect of l ateral turbulence intensity on drop diameter observed in FCMP da ta

PAGE 90

90 Wind Velocity and Turbulence Intensity Dependency of the Raindrop Size Distribution The influence of wind on the RSD was investigated by stratifying the data statistics in to different wind speed and turbulence intensity regimes. The gamma parameters describing the RSD and the mean wind speed ( ) and turbulence intensities ( and ) were calculated for each one minute segment. The dependency of the three par ameters on the wind velocity and turbulence intensities is shown in Figure 4 9 and Figure 4 10 These data indicate that and remained constant with increasing or and that the variance of each of the variables decreased with increasing Figure 4 9 and Figure 4 10 also demonstrates that the variability o f and decreased with increasing R. The significance is that while a variability of drop diameter was observed in multiple and regimes, the RSD was independent of and Another observation was ma de regarding and A comparison of the mean values of and approximately 1.2 and 1.3 mm 1 respectively was significantly different than 3.6 and 3.4 mm 1 observed from typhoon data collected by Chang et al. (20 09) ; however, Figure 4 12 demonstrates that the relationship between the measured and is similar to the empirical relationship established by Zhang et al. (2003) ( 4 3 ( 4 3 ) Possible sources for the differences in mean values are that: (1) the RSD data gathered by Chang were observed in wind speeds < 8.0 m/s, (2) there is less noise in the data when implementing an articulating instrument platform, or (3) there are too few one minute measurements for a reasonable comparison. Presently the reason for the dissimilarity is unclear and more data would be n ecessary to make an assessment.

PAGE 91

91 Figure 4 9 VORTEX2 gamma parameters observed in multiple wind conditions and rainfall intensities

PAGE 92

92 Figure 4 10 FCMP Hur ricane Ike and Irene gamma parameters observed in multiple wind conditions and rainfall intensities

PAGE 93

93 Figure 4 11 FCMP and VORTEX2 gamma parameters observed in multiple wind conditions and rainfall intensit ies

PAGE 94

94 Figure 4 12 Observed shape slope relation

PAGE 95

95 Comparison of R aindrop S ize D istribution Models to Measured R aindrop S ize D istribution Data in Multiple Wind Velocities RSD models serve as a simple method f or acquir ing an adequate RSD for WDR models ; thus, one of the thrusts of this project was to validate the application of these models in extreme WDR scenarios. The models include the Marshall and Palmer (1948), Best (1950), and Willis and Tattleman (1989) dependent RSD models and the three parameter gamma model using the mean of the parameters calculated from the gathered data ( and ). VORTEX2 and FCMP Hurricane Ike data were stratified into two (< 15 and > 15 m/s), two (< 0.2 and > 0.2), two (< 0.2 and > 0.2), and five (0 20, 20 40, 40 60, 60 80, and 80+ mm/hr) regimes illustrated in Figure 4 13 Figure 4 15 These figures demonstrate qualitatively that there is no apparent relationship between the RSD and or and that throughout all regimes the Best model is most accurate. To validate these observations, mean s quare error ( ) values were calculated and listed in Figure 4 16 Figure 4 18 Each value was computed as f o l l o w s : ( 4 4 ) where is the measured RSD and is the fit RSD. The values confirm that there was no significant di fference in the performance of the dependant models below or above 15 m/s, below or above 0.2 or below or above 0. 2 values also verify that the Best model is the most accurate, yielding the least error.

PAGE 96

96 Figure 4 13 Model raindrop size distribution and measured raindrop size distribution comparisons (N indicates the number of averaged records)

PAGE 97

97 Figure 4 14 Model raindrop si ze distribution and measured raindrop size distribution comparisons (N indicates the number of averaged records)

PAGE 98

98 Figure 4 15 Model raindrop size distribution and measured raindrop size distribution compari sons (N indicates the number of averaged records)

PAGE 99

99 Mean Square Error Values Model Marshall and Palmer Best Willis and Tattleman Gamma VORTEX2 FCMP VORTEX2 FCMP VORTEX2 FCMP VORTEX2 FCMP U (m/s) U (m/s) U (m/s) U (m/s) U (m/s) U (m/s) U (m/s) U (m/s) Rainfall (mm/hr) < 15 > 15 < 15 > 15 < 15 > 15 < 15 > 15 < 15 > 15 < 15 > 15 < 15 > 15 < 15 > 15 0 20 2.03 1.14 3.69 2.64 0.31 0.25 0.58 0.29 1.88 1.09 3.20 2.33 0.98 0.49 0.23 0.26 20 40 1.14 1.54 1.43 1.74 0.04 0.26 0.10 0.04 1.07 1.55 1.41 1.56 0.11 0.29 0.53 0.65 40 60 1.07 0.99 1.77 1.78 0.03 0.19 0.21 0.08 1.00 1.05 1.77 1.60 0.07 0.15 0.53 0.71 60 80 1.00 1.09 1.99 1.83 0.04 0.23 0.27 0.10 0.94 1.00 2.03 1.73 0.07 0.08 0.50 0.73 80 + 1.11 1.21 2.25 2.39 0.08 0.19 0.10 0.05 1.10 1. 22 2.30 2.34 0.17 0.15 0.87 0.84 Mean 1.27 1.19 2.23 2.08 0.10 0.22 0.25 0.11 1.20 1.18 2.14 1.91 0.28 0.23 0.53 0.64 Model Mean 1.69 0.17 1.61 0.42 Figure 4 16 Mean square error values of raindrop size d istribution models stratified by U and R Mean Square Error Values Model Marshall and Palmer Best Willis and Tattleman Gamma VORTEX2 FCMP VORTEX2 FCMP VORTEX2 FCMP VORTEX2 FCMP TI U TI U TI U TI U TI U TI U TI U TI U Rainfall (mm/hr) < 0 20 > 0 20 < 0 20 > 0 20 < 0 20 > 0 20 < 0 20 > 0 20 < 0 20 > 0 20 < 0 20 > 0 20 < 0 20 > 0 20 < 0 20 > 0 20 0 20 2.14 1.94 3.40 3.27 0.36 0.28 0.49 0.41 1.99 1.79 2.93 2.91 1.06 0.92 0.23 0.20 20 40 1.16 1.13 1.58 1.53 0.05 0.04 0.03 0.06 1.10 1.05 1.47 1.46 0.11 0.11 0.63 0.58 40 60 1.07 1.09 1.70 1.75 0.05 0.03 0.09 0.16 1.01 1.01 1.59 1.73 0.07 0.08 0.69 0.60 60 80 1.04 0.98 1.79 1.91 0.06 0.04 0.13 0.20 0.98 0.91 1.74 1.91 0.07 0.08 0.70 0.65 80 + 1.17 1.06 2.36 2.34 0.10 0.08 0.05 0.07 1.18 1.04 2.33 2. 34 0.16 0.18 0.87 0.82 Mean 1.32 1.24 2.17 2.16 0.12 0.09 0.16 0.18 1.25 1.16 2.01 2.07 0.29 0.27 0.62 0.57 Model Mean 1.72 0.14 1.62 0.44 Figure 4 17 Mean square error values of raindrop size distributio n models stratified by TI U and R

PAGE 100

100 Mean Square Error Values Model Marshall and Palmer Best Willis and Tattleman Gamma VORTEX2 FCMP VORTEX2 FCMP VORTEX2 FCMP VORTEX2 FCMP TI V TI V TI V TI V TI V TI V TI V TI V Rainfall (mm/hr) < 0 20 > 0 20 < 0 20 > 0 2 0 < 0 20 > 0 20 < 0 20 > 0 20 < 0 20 > 0 20 < 0 20 > 0 20 < 0 20 > 0 20 < 0 20 > 0 20 0 20 2.05 1.99 3.40 4.59 0.32 0.31 0.49 0.91 1.90 1.84 2.94 4.28 1.00 0.95 0.22 0.43 20 40 1.19 1.09 1.56 1.48 0.04 0.04 0.04 0.17 1.13 1.01 1.46 1.64 0.12 0.10 0.6 2 0.36 40 60 1.12 1.00 1.71 1.84 0.04 0.03 0.11 0.30 1.06 0.92 1.61 2.21 0.08 0.06 0.67 0.31 60 80 1.01 0.99 1.80 0.05 0.03 0.14 0.96 0.91 1.77 0.07 0.08 0.68 80 + 1.16 1.02 2.34 2.17 0.08 0.10 0.05 0.38 1.16 0.98 2.33 2.93 0.16 0.18 0.86 0. 63 Mean 1.31 1.22 2.16 2.52 0.11 0.10 0.17 0.44 1.24 1.13 2.02 2.76 0.29 0.28 0.61 0.43 Model Mean 1.80 0.20 1.79 0.40 Figure 4 18 Mean square error values of raindrop size distribution models stratified by TI V and R

PAGE 101

101 Peak to Mean Ratio of Rainfall Intensities Currently, the widely accepted standard used to estimate the design wetting rate on building faades (BSI EN ISO 15927 3) is based on a minimum of ten years of hourly and data. The design fre e stream WDR ( ) is calculated as the 67th percentile of the values determined from 4 5 (based on Lacy, 1965): ( 4 5 ) where is the hourly mean wind speed, is the hourly rainfall total, is the hourly mean wind direction from north, and is the wall orientation relative to north. Designing for extreme WDR events may require a more str ingent method due to the stochastic nature of both wind and rain in time scales less than one hour. Figure 4 20 illustrates the peak to mean ratios of FCMP Hurricane Ike data from one minute to one hour (Durst, 1960). This figu re demonstrates that designing for a shorter duration peak can significantly increase the design For instance, using hourly mean FCMP Hurricane Ike data the ; however, if the 10 minute peak is used, The peak to mean ratios of can be useful in determining the WDR load on building components; thus, the effect of wind and precipitation type on the peak to mean ratios was investigated. Figure 4 19 and Figure 4 20 illustrate the peak to mean ratios of VORTEX2 and FCMP Hurricane Ike data, respectively. VORTEX2 data only contains peak to mean ratios beyond ten seconds (due to the temporal resolution of the instrumentation) and under 15 m/s. In both figures, it is clear that peak to mean ratios observed in convective precipitation are lower. To verify this observation, a two sample

PAGE 102

102 t test was conducted to compare the datasets at one and ten seconds (VORTEX2 a nd FCMP data, respectively) from convective and stratiform precipitation. The t test indicated that difference of the peak to mean ratios in different precipitations types are statistically significant at the 95% significance level. Figure 4 20 also indicates that there are lower peak to mean ratios in U > 15 m/s. A t test was conducted and indicated that the difference of peak to mean ratios in < 15 m/s and > 15 m/s is statistically significant at the 95% significance level. The significance of different peak to mean ratios could ultimately lead to design that are orders of magnitude different. The applicability of a peak to mean ratio would then depend on site specific probabilities of type of precipitation and wind speed. The study of peak to mean ratios of rainfall intensities could yield results that are useful in design practice. Table 4 2 Peak to mean ratios of U and R Peak Duration (sec) U Peak to Mean Ratio (Durst, 1960) R Peak to Mean Ratio Multiplier of 1 hr Value 3 1 5 50 0 75 0 60 1 2 7 0 8 4 300 1 1 3 5 3 9 600 1 1 2 6 2 9 1800 1 0 1 5 1 5 3600 1. 0 1. 0 1. 0

PAGE 103

103 Figure 4 19 Peak to mean ratios of VORTEX2 data (N indicates number of one minute segments)

PAGE 104

104 Figure 4 20 Pea k to mean ratios of FCMP data (N indicates number of one minute segments)

PAGE 105

105 Comparison of Ground Measured Rainfall Intensity and Estimated Reflectivity to Weather Surveillance WSR 88D Estimated Rainfall Intensity and Measured Reflectivity FCMP Hurricane Ike measurements of rainfall intensity ( ) and estimates of reflectivity ( ) were compared with remotely sensed estimates of and measurements of reflectivity from the National Weather Service WSR 88D Doppler Radar KHGX. WSR 88D Doppler radar data consist ed of reflectivity measurements and rainfall estimates at locations along a radial grid extending from the radar location (~ 40km). A linear interpolation of the data from the four nearest grid locations was performed to estimate both the rainfall intensit y and reflectivity of the cell above the disdrometer (elevation from 478 m to 540 m). The temporal resolution of the data was approximately five minutes therefore the collected data was divided into five minute segmental averages. VORTEX2 data was not comp ared to WSR 88D Doppler Radar data due to the short durations of each of the 22 records (<10 minutes). The comparison between the disdrometer estimated reflectivity and radar measured reflectivity is illustrated in Figure 4 21 Generally, there is good agreement between the two. The mean absolute error, 7. 6 dBZ, calculated as : ( 4 6 ) where is the number of observations, is the radar measured reflectivity, and is the disdrometer estimated reflectivity, confirms this observation. Rainfall intensities measured by the PIP and estimated by radar are compared in Figure 4 22 This figure shows that there is relatively g ood agreement below 10 mm/hr. Above 10 mm/hr the default relationship, employed at KHGX, appears to underestimate the observed Mean absolute errors below and above the 10 mm/hr

PAGE 106

106 regimes are 2.8 mm/hr and 23.3 mm/hr, respectively. Figure 4 23 demonstrates the relationship between the measured ground level reflectivites and rainfall intensities, and the best fit curve. Figure 4 24 confirms that the default Z R model a nd the best fit curve agree reasonably well below 10 mm/hr. Commonly used empirical Z R relationships, as recommended by the WSR 88D Operational Support Facility, are found in Table 4 3 The Rosenfeld tropical rela tionship was the best predictor for the ground level estimated reflectivity, with an value of 0.58. For the range of reflectivity values observed in this storm, all relationships underestimated the observed rainfall intensity ( Figure 4 24 ). Table 4 3 Comparison of Z R models Model Z Z ( dBZ ) R 2 Best Fit Curve Rosenfeld tropical Default WSR 88D Marshall/Palmer Figure 4 21 Comparison of disdrometer estimated an d radar measured reflectivit y

PAGE 107

107 Figure 4 22 Comparison of disdrometer measured and radar estimated rainfall intensit y Figure 4 23 Observed Z R relationship

PAGE 108

108 Figure 4 24 Comparison of observed and recommended Z R relationships Summary This chapter presented the analysis and results from data collected during the VORTEX2 and FCMP field campaigns. Results from th e analysis proved that while a relationship between drop diameter and wind was observed, the RSD was unaffected by wind. It was discovered that the Best model (1950) was the model that best fit the observed data. Obser vations were also made regarding the application of one hour data in WDR design as well as commonly used empirical Z R relationships. The next chapter applies the findings from this analysis to the design and development of a WDR simulation system for the water penetration resistance evaluation of low rise buildings in a full scale wind tunnel.

PAGE 109

109 CHAPTER 5 DEVELOPMENT OF A W IND DRIVEN RAIN SIMULATION SYSTEM FOR THE WATER PENETRATION RESISTANCE EVALUATION OF LOW RISE BUILDINGS IN A FULL SCALE WIND TUNNEL This chapter presents a new method to design a wind driven rain (WDR) simulation system to achieve a prescribed rain deposition rate on a low rise structure in a full scale wind tunnel test facility. The method was successfully applied during the commissioning of the Full Scale Test Facility at the Insurance Institute for Business & Home Safety (IBHS) Research Center. Essentially, the facility is a wind tunnel with a cross section that is large enough to test a full size two story building. IBHS requested guidance on t he addition of a water injection system to simulate WDR in the test chamber. The water injection system consists of a grid of spray nozzles located at the entrance of the test chamber. The primary design objective was to achieve 203 mm/hr rain deposition r ate on the windward wall of a single story building located 10 m downwind of the spray system. The second objective was to select a conventional, inexpensive spray nozzle that produced a raindrop size distribution (RSD) representative of measurements in tr opical cyclones and supercell thunderstorms. The Best (1950) model was selected for this application, based on the results presented in Chapter 4. Research was carried out in four stages. First, the wetting uniformity of a single nozzle was evaluated to de termine what level of resolution could be achieved in the system design. It was determined that while characterizing the RSD of the entire spray from a single nozzle was possible, mapping out the trajectory of the individual particle sizes emitted from the nozzle (in the statistical sense) over the spray cone was not easily achieved. Second, disdrometer measurements were performed in stagnant air

PAGE 110

110 and in steady wind to determine if the presence of wind caused the changes in the RSD. The differences were foun d to be inconsequential for this application. Based on this result, spray nozzles were tested in stagnant air conditions in the third stage. Twelve commercially available hydraulic spray nozzles were evaluated in a large test stand with the nozzle spraying perpendicular to the floor. RSD measurements were made along the radial extents of the area receiving spray and integrated using the method of circular disks to characterize the RSD for the entire spray. These results were used to select the best nozzle. Finally, validation tests were performed in the IBHS Research Center. The remainder of this chapter is organized into five sections. The first section discusses the design specifications for the rain simulator. The remaining sections present information a bout the four research stages. Design Specifications for the Rain Simulator in the IBHS Research Center IBHS specified that the rain deposition rate on the windward wall of the test building should be of 203 mm/hr in 58 m/s winds. The decision to use 203 m m/hr was based on duplicating current water penetration resistance test standards, including ASTM E331 00, ASTM E547 00, ASTM E1105 05, and ASTM E2268 04. The deposition rate is not the same as the wind driven and horizontal rain intensities; therefore it is not easy to relate the specified value to an actual rain event. More importantly, the discharge rate of the nozzles can be directly inferred from deposition rate. This section discusses these issues in detail. is commonly presented in several forms. The flux of rain falling vertically is defined by the horizontal rainfall intensity ( ) and has units of LT 1. It is equal to the volume of water collected over a specified duration

PAGE 111

111 divided by the horizontal a rea of the collection chamber. WDR intensity ( ) is the flux of rain passing through a vertical plane. scales with wind speed (Eq. 2 22), therefore it is significantly larger than in strong winds. The third term is the rain de position rate ( ), which is the specified wetting rate on the building faade (203 mm/hr). It differs from because of the flow structure interaction. Only a fraction of the wind driven rain wets the wall. Table 5 1 presents values for and from multiple field studies and the value used in ASTM water penetration resistance test standards. Th e non italicized values in the table are the values reported in the literature. The italicized values are estimates. and were converted using the Lacy (1967; Eq. 2 2 4 ) equation. and were converted following BS EN ISO 15927 3:2009 (Eq. 2 26) assuming open exposure terrain, flat topography and no obstructions immediately upwind. The wall factor ( ) was selected from Table 4 in the standard, which contains values for a wide range of building shapes. was select ed based on its frequency as a windward wall coefficient. Of the three intensity parameters, only can be directly modulated in the IBHS Research Center. can be determined rationally by dividing the discharge rate of a Table 5 1 Rainfall Intensities All values shown as mm/hr. RWDR was computed a ssuming the wind speed U = 60 m/s. Italicized values are estimates. Source R H R WDR R F = W R WDR Hurricane Ike Mean (FCMP) 6 48 19 Typical Cat 3 5 Peak (Lonfat et al., 2004) 12 95 38 2004 2006 Tropical Cyclones (Tokay et al., 2008) 19 147 59 NOAA TP40 100 year 6 hr Event (NWS, 1961) 42 333 133 ASTM Test Standards 64 509 203 Supercell T storm peak (Smith et al., 2000) 300 2377 951

PAGE 112

112 single noz zle (L 3 T 1 ) by the grid cell area (L 2 ), which was nominally 660 mm x 610 mm. Prior research (e.g. Choi, 1994; Blocken and Carmeliet, 2002) has shown that does not vary significantly with wind speed. Therefore, the major challenge in this research was to select a spray nozzle that would best replicate a naturally occurring RSD, therefore resulting in a value approximately equal to 0.4. A nozzle with too many small drops would cause a decrease in the parameter, and vice versa. Spray Uniformity an d the Effect of Wind Velocity on the Raindrop Size Distribution Manufacturer supplied specifications for hydraulic spray nozzles typically include the discharge rate, spray angle and spray pattern (circular or square). The RSD is not given for off the shel f products, such as the nozzles considered in this study. Therefore it was necessary to devise a test protocol to characterize the RSD of each nozzle. Given the absence of information about the RSD, several questions needed to be answered before this proto col could be developed: 1. Are the wetting rate and RSD uniform along the radial extent of the spray pattern? Would one measurement suffice (and thereby save significant time and expense of testing) or are multiple measurements required to develop a composite RSD for the entire spray region? 2. How does the presence of wind change the RSD? Can the RSD characterization procedure be performed in still air, eliminating the need for expensive and time consuming tests? To answer question 1, a simple experiment was per formed to quantify the variability of the spray using the apparatus shown in Figure 5 1 a. A single nozzle was aligned perpendicularly to the center of a vertically oriented collection chamber with 12 receivers. Ea ch partition drained into its own graduated cylinder. The test consisted of (1) covering the chamber, (2) setting the flow rate of the nozzle using a pressure regulator, (3) removing the cover to allow water to enter the receivers, (4) covering the

PAGE 113

113 chamber after five minutes, (5) shutting off the nozzle and (6) reading the graduated cylinder and recording the accumulated water volume. Ten nozzles were evaluated in this manner, and the results of all tests clearly demonstrated a non uniform wetting pattern. The most uniform spray pattern observed in test series is shown in Figure 5 1 B. The wetting rates varied by almost a factor of three (123 301 mm/hr) with a coefficient of variation, CoV, of 0.25. The implication is that multiple measurements inside the spray region would be required. This led to the radial profiling technique discussed in the next section. (a) Collection Chamber (b) Sample results for a BETE WL 1 1/2 nozzle Figure 5 1 Apparatus to measure the spray uniformity of a single nozzle (photo courtesy of author) To answer question 2, a series of tests were performed in stagnant air and in a steady jet generated by the wind generator at UF. The e xperimental configuration consisted of an OTT PARSIVEL and DMT PIP located 3 m downwind of a 3 m x 3 m uniform, open jet. The sensors were oriented horizontally such that the jet and laser planes were parallel. Measurements were taken for wind velocity U = 8.9 m/s, 17.9 m/s, and 35.8 m/s and compared to measurements taken in still air. The still air test was

PAGE 114

114 performed with the nozzle facing downward and the PARSIVEL in an upright (standard) position and located at the center of the circular spray area. Thre e tests were performed for each case. The sampling duration was five minutes. Results are shown in Figure 5 2 The vertical axis is the number concentration which is the number of particles divided by a sam ple volume and diameter (dependent on wind speed, Equation 2 28). The horizontal axis is the raindrop size diameter. The wind speed and wind driven rain intensity are shown in the legend. The figure demonstrates that regardless of wind speed, the RSDs i n the < 1 mm regime match the stagnant air case for both instruments. Concentrations for drop diameters greater than 1 mm were greater than the stagnant air case; however, this behavior was expected and attributed to the narrowing of the spray cone in the presence of wind. As the following section will explain, all of the nozzles evaluated in stagnant air, ejected large droplets ( ) towards the outer edge of the spray area. Thus, as the spray cone narrowed, a higher concentration of larger drops was expected. The variation of the wind driven rain intensities was attributed to the small sample volumes resulting from five minute tests ). Mueller and Sims (1966) determined that a minimum sample volume of i s necessary to get re liable estimates of the rainfall intensity and reflectivity (within 10% at a 95% confidence interval). The required test time to achieve this sample volume (employing PARSIVEL and PIP disdrometers) is a minimum of 15 minutes; however, this test duration wa s time and cost prohibitive given the fuel, water, and labor resources necessary to operate the wind generator.

PAGE 115

115 The results from these tests indicate that the RSD behavior for the majority of the drops (i.e., less than 1 mm) was unaffected by wind and the RSD of larger drops behaved as expected. Therefore, an investigation of the characteristic nozzle RSD in stagnant air was acceptable. Figure 5 2 Comparison of raindrop size distribution s measured in stagn ant air and in a steady wind by the PARSIVEL (left) and PIP ( right, BETE WL 1 1/2 at 138 kPa) Characterization of the R aindrop S ize D istribution of a Spay Nozzles in Stagnant Air: A Proxy for Full Scale Testing Specimen Matrix The 12 full cone hydraulic no zzles shown in Table 5 2 were selected for evaluation based on their discharge rates and cost (> 450 nozzles are used in the IBHS Research Center Full Scale Test Facility). Table 5 2 Spray n ozzles e valuated in thi s s tudy BETE Lechler Steinen Spray Systems WL 1 460.648.30.BE SM 151W 1/8GG 8W WL 2 460.728.30.BE SM 303W 3/8GG 17W WL 3 460.768.30.BE TF 8 TF 10

PAGE 116

116 Experimental Configuration The experimental setup consisted of a 4.9 m x 4.9 m x 4.9 m chamber with enclosed sides to reduce the effect of ambient wind. The nozzle was suspended 3.1 m (10 m) above the ground in the center of the chamber facing downward. Four radial extents were marked on the floor in 0.3 m increments, as shown in Figure 5 3 The PARISIVELs were relocated to each position for five minutes of data capture. Figure 5 3 PARSIVEL t est locations for nozzle characterization Analysis The PARSIVEL outputs drop counts in a 32 x 32 matrix. Each cell corresponds to a specific combination of drop diameter and fall velocity (see Chapter 2 for a detailed explanation). Values were summed for all velocities for each diameter bin, and then averaged over the four radial extents, yielding a 32 x 1 array of drop counts per diameter along the radial extent. Results for one nozzle (BETE WL 3) are shown in

PAGE 117

117 Figure 5 4 for D = 0.31 mm to D = 3.8 m. The black lines correspond to the number of drops as a function of radial distance from the centerline of the nozzle. The dashed lines correspond to the maximum theoretical distance the particles should travel based on the spray angl e of the nozzle and the initial velocity. The governing equations for the trajectory of a smooth rigid sphere acted upon by gravity and wind are Reynolds (Re) number dependent. For Stokes flow conditions (Re < 1), the force acting on the raindrop is equal to: ( 5 1 ) For 1
PAGE 118

118 ( 5 6 ) ( 5 7 ) ( 5 8 ) The initial velocities, i.e. were determined using a Phantom V9.1 high speed camera recording 1152 x 720 pixel images at 1000 fps and passed through an edge filter ( Figure 5 5 ). It is evident from Figure 5 4 that the smaller particles tend to fall closer to the centerline of the nozzle. The outermost fall locations match the theoretical estimates for D = 0.68 mm to 1.68 mm. The particles that fell outside of this region exceeded the limit due to the temporary walls not completely preventing air currents from forming inside of the test chamber and water drops generating turbulence during their descent. For the smaller drops, wind velocities as low as 0.1 m/s are sufficient to cause the deviation observed in the data. The larger drops did not reach the theoretical limit and tended to fall further from the center of the spray area. This behavior was observed visually in all of the nozzles and is hypothesized to be a result of the higher momentum of the larger drops. The helical ducting in the nozzles (all of the nozzles had similar des igns) introduces an angular velocity component into the flow; consequently, larger drops (higher mass) have a higher tangential momentum and are ejected farther from the center. Plots of the other nozzles (see appendix) demonstrate that the spray pattern f or all nozzles was similar.

PAGE 119

119 Figure 5 4 Count of drops radially outward from nozzle centerline

PAGE 120

120 Figure 5 5 Determining initial velocity using high speed fo otage (photo courtesy of author) To compute the number of drops per diameter for the entire spray region from the data averaged over the radial extent, each array was integrated using the method of circular discs: ( 5 9 ) where is the total number of drops of diameter is the radial distance from the center to the measurement location, is the number of observed diameter droplets at location and is the sample area of the PARSIVEL. The nozzle flow rate was also computed from the PARISVEL data to verify mass continuity: ( 5 10 ) where is the flow rate, and is the sample time. The estimated values agreed very well with the known discharge values. Errors were on the order of 15% or less. The RSD of each nozzle was calculated from:

PAGE 121

121 ( 5 11 ) where is the circular spray area and is the terminal velocity of a diameter drop. Figure 5 6 shows the calculated RSD for all of the nozzles (RSDs of other nozzles are independently listed in the appendix). As the figure shows, the major difference in the RSD of the nozzles was the number concentration of large particles. The concentration of smaller particles was virtually identical for raindrop diameters <= 1.0 mm. Therefore the selection of the BETE WL3 was made based on achieving the largest number of large drops in the flow. Figure 5 6 R aindrop size distribution of BETE WL3 in stagnant air conditions

PAGE 122

122 Validation of the W ind D riven R ain Simulation System at the IBHS Research Facility Upon selection of the nozzle, the wind driven rain simulation system was constructed. The system consists of three nozzle grids for each of the three horizontal fan cells. The flow rate and pressure of each grid is actively controlled by automated gate valves. Horizontal nozzle spacing was limited to either 1320 mm or 66 mm due to the 1320 mm spacing of the air foils. It was observed from testing at UF that the shorter horizontal spacing would yield a more uniform spray, particularly in high wind speeds; thus, 660 mm was selected. Similarly, shorter vertical spacing would yield more uniform spray patterns; however, the closer the spacing, the lower the required pressure and flow rate per nozzle. The lowest pressure recommended by the manufacturers was 34.5 kPa; thus, the minimum vertical spacing required was 610 mm. To validate the design recommendations a PARSIVEL and PIP were used to take RSD measurements at multiple locations in the stream. The instruments were placed on vertical steel lattices at 10 m and 3 m from the air foils ( Figure 5 7 ). Flow rates were determined to be approximately 3.8 L/min per nozzle (using inline digital flow gauges, corresponds to approximately 34.5 kPa). Five minute measurements of the RSD were taken at 2.0 m, 3.0 m, 4.0 m, 5.2 m, and 6.0 m for each wind speed. The measured RSDs matched the RSD predicted by the Best (1950) model at an intensity of 509 mm/hr, confirming the choice of nozzle and spacing ( Figure 5 8 ). The mean of all the RSDs also agreed with the Best (1950) model ( Figure 5 9 ). This observation implies that for full scale simulations, the wetting rate reaching the building faade will be approximately 203 mm/hr and the RSD will be representative of natural conditions.

PAGE 123

123 Figure 5 7 Instrument arrangement at the I nsurance Institute for Business & Home Safety Research Center (photo courtesy of author)

PAGE 124

124 Figure 5 8 Measured raindrop size d istribution s at multiple heights and wind velocities

PAGE 125

125 Figure 5 9 Comparison of measure d raindrop size distribution s and Best (1950 ) model Summary This chapter presented the design of a realistic wind driv en rain simulation system that can reproduce the rainfall intensity and drop size distributions obtained from the field measurement activities (Chapter 4). The simulation system was implemented at the Insurance Institute for Business & Home Safety Researc h Facility for the water penetration resistance evaluation of low rise buildings. The following chapter will summarize contributions and conclusions of the research described in this document and present recommendations for future studies.

PAGE 126

126 CHAPTER 6 SUMMARY, CO NCLUSIONS,AND RECOMMENDATIONS This chapter summarizes the efforts taken to advance the knowledge base by characterizing extreme wind driven rain (WDR) events, creating and evaluating a WDR measurement technique, and the design of a full scale WDR simulatio n system. Characterization of Extreme W ind D riven R ain Events The VORTEX2 and FCMP field campaigns yielded 17 thunderstorm and two tropical cyclone datasets. For these activities, a novel approach was taken to minimize errors associated with strong winds. The instrumentation platforms continuously align the disdrometers towards the mean rain vector using continuous wind velocity and direction feedback. The first instrument platform was designed to be mounted on one of the ruggedized FCMP meteorological m easurement stations to record rain data at 3m and wind data at 5m; however, the employed disdrometer was cost prohibitive when considering a multiple instrument array. Thus, as part of this project, a low cost, easy to handle, robust instrument platform w as designed and constructed.To achieve a low cost solution (20% of the cost of the high performance disdrometer) OTT PARSIVELs were employed; however, the OTT PARSIVELs have a lower resolution and sampling rate than the DMT PIP. Moreover, the disdrometer is not intended to be used in strong winds. Thus, three laboratory experiments were conducted to investigate the errors exhibited by the PARSIVEL associated with strong winds. Proof of Concept The first two experiments were measurements of the raindrop siz e distribution (RSD) emitted from a nozzle in stagnant air conditions. One test compared the measured RSD of the PARSIVEL to the RSD determined by a morphological image

PAGE 127

127 processing algorithm that was created to count the number of droplets captured in an o il medium. Results demonstrated that for practical purposes, when the PARSIVEL is oriented towards the mean rain direction, the measured RSDs compared favorably to those measured by the image processing algorithm. The second test evaluated the effect of oblique trajectory angles on the measured diameter and velocities. High tolerance steel bearings, of multiple diameters, were dropped through the laser band of the PARSIVEL and different angles. The different angles replicated the effect of wind on drop lets passing through the measurement area of a stationary instrument (i.e. laser band continuously parallel to the ground). These tests revealed that at oblique trajectory angles, diameters of drops are measured with high fidelity; however, the measured ve locities are unrealistically reduced. The consequence is unrealistic large number concentrations ( ). The third laboratory experiment compared measurements from the PARSIVEL, oriented at different angles, and a DMT PIP mounted on a gantry system ex posed to multiple wind speeds and rainfall intensities. Results from these tests confirmed the findings from the stagnant air tests; the PARSIVELs exhibited erroneous measurements resulting from incorrect orientation which are magnified by higher wind vel ocities. These observations validated the concept taken in the design of the articulating instrumentation platforms. The articulating instrumentation platforms continuously aligned the disdrometers such that the laser plane was perpendicular to the mean rain vector. Field data from collocated stationary and articulating instrumentation platforms confirm that in moderately high wind speeds (> 10 m/s), data from the stationary

PAGE 128

128 instrumentation becomes compromised ; while the data from the articulating instr umentation exhibited no wind induced error. Conclusions from Field Data Results The simultaneous wind and rain data gathered by the articulating instruments, to stream R SDs in extreme wind events. Thus, as part of the research project the following five topics were investigated:1) the effect of wind on the drop diameter, 2) the effect of wind on the RSD, 3) comparison of RSD Models to RSDs measured in multiple wind veloci ties, 4) calculation of the peak to mean ratios of and 5) comparison of rain data measured in extreme events to WSR 88D data. Effect of wind velocity and turbulence intensity on raindrop d iameter The effect of wind velocity and turbulence intensity on raindrop size was investigated using data from both field campaigns. Two main trends were observed: 1) mean drop diameter tends to increase with increased wind velocity and 2) mean drop diameter tends to decrease in the presence of high turbulence intensities. The cause of these trends is postulated to be attributable to one or both of the following possibilities: 1) It is possible that in increased wind velocities the high drag forces acting on the droplets overcome the surface tension holding the shape together and breakup occurs or 2) high turbulence in tensities cause the probability of coalescence to increase. These conclusions are intended to motivate further study. This is a limited dataset and more data is necessary for a comprehensive study to be performed and definite conclusions to be made.

PAGE 129

129 wind velocity and turbulence intensity dependency of the raindrop size distribution Observations of drop diameter changing in the presence of multiple wind scenarios led to investigate the effect of wind on the RSD. For this investigation, wind velocity, turbu lence intensities, and the gamma parameters for each one minute segment of VORTEX2 and FCMP data were calculated. These data ( Figure 4 11 ) suggest that and remain constant with increasing longitudinal velocity, and longitudinal and lateral turbulence intensities; however, the variance of each of the variables decreased with increasing longitudinal velocity. Thus, it was concluded that that the RSD does not significantly change with increased wind speed, or longitudin al and lateral turbulence intensities. When investigating the relationship between and (Figure 5 12) it was observed that the measured relationship differs from the empirical relationship established by Zhang (2003, Equation 5 12). A possible source for the difference is that there are too few one minute measurements. If the obser ved relationship is valid, the implication could ultimately mean incorrect interpretation of RSD parameters from reflectivity measurements. Presently, the reason for the dissimilarity is unclear and more data is necessary to make an assessment. Comparison of Measured Data to R aindrop S ize D istribution Models RSD models are used by the wind engineering community as a simple technique to acquire an adequate RSD for WDR modeling. Thus, an evaluation of the applicability of current RSD models to data gathered in strong winds was performed. Field measured data was compared to the Marshall and Palmer (1948), Best (1950), and Willis and Tattleman (1989) dependent models and the gamma model using the mean

PAGE 130

130 parameters from the measured data. The results indicate that the difference in the performance of the models below or above 15 m/s, below or above 0.2 or below or above 0.2 (Figures 5 18 5 20) was insignificant. The data also shows that the Best (1950) model yields the least error and is thus the most accurate. Comparison of Measured Data to WSR 88D Data FCMP Hurricane Ike data were compared with remotely sensed estimates from the National Weather Service WSR 88D Doppler Radar KHGX. Generally, there was good agreement between disdrometer estimated and radar measured reflectivities. The data shows that there is also relatively good agreement between the ground level measured rainfall intensity and the radar estimated rainfall intensity below 10 mm/hr. Above 10 mm/hr the default Z R relat ionship, employed at KHGX, underestimated the observed rainfall intensity. Thus, other commonly used empirical Z R relationships (Table 5.3, as recommended by the WSR 88D Operational Support Facility) were compared. Of the recommended relationships, the R osenfeld tropical relationship more closely resembled the observed data; however, all of the Z R relationships underestimated the observed rainfall. Peak to Mean Ratio of Rainfall Intensities Peak to mean ratios of rainfall intensities were calculated for their potential use in design practice. Currently, the widely accepted standard (BSI EN ISO 15927 3) calculates the rain deposition rate from hourly and data. The research demonstrates that designing for a shorter duration peak can significantly in crease the design wind driven rainfall intensity ( ) and subsequently, the deposition rate ( ). For instance, using 10 minute FCMP Hurricane Ike data, the design free stream wind driven rain intensity is approximately three times larger th an using hourly data. For

PAGE 131

131 design, the applicability of a peak to mean ratio would depend on site specific probabilities of type of precipitation and wind speed. The presented research shows the potential value of peak to mean ratios of rainfall intensitie s in design practice; however, more data should be collected to assess the validity of the peak to mean ratio curves presented herein. Design and Implementation of a Full Scale W ind D riven R ain System The Insurance Institute for Business & Home Safety ta sked researchers at the University of Florida with the design of a WDR simulation system for the IBHS Research Center Full Scale Test Facility. The design was based on field measured RSD data and intended to reproduce the rain deposition rate specified in current test standards. The and RSD, selecting the most appropriate nozzle, and validating the full scale WDR simulation system. Nozzle Characterization The OTT PARSIVEL was utilized to investigate the characteristics of nozzles te the approximate necessary flow rate. Measurements in the presence of wind revealed that there was negligible change in the RSD indicating that the particles accelerated with the flow with little interaction. Thus, it was deemed acceptable to characteri ze the RSD of individual nozzles in stagnant air. Of the tested nozzles, the BETE WL 3 exhibited an RSD that closely resembled the Best (1950) RSD model (the model that best fit field measured data). Nozzle spacing was set to the minimum possible spacing g iven space and flow rate limitations. Full Scale Implementation Upon determining the spray system characteristics, the rain simulation system was constructed at the IBHS Research Center. Free stream WDR measurements were

PAGE 132

132 taken at the facility with the PA RSIVEL and the PIP at multiple locations in the stream. The measurements indicate that the RSD closely resembles the RSD predicted by the Best (1950) model. Thus, future full scale simulations will be representative of natural conditions. Recommendations f or Future Research Recommendations for Instrumentation Construction of more instrument platforms will facilitate the investigation of the spatial correlation of WDR. The instruments can also be easily adapted to make tower mounted meteorological observat ions. Thus, the vertical and horizontal evolution of WDR can be investigated. Currently the design of the articulating instrumentation platforms allows for the upgrade of real time data streaming. The availability of real time data streaming may aid age ncies such as meteorological institutions in updating forecasts, emergency management agencies in assignment of resources during and after extreme events, and FCMP teams responding to errors by instrumentation during data acquisition. Future research, util izing the articulating instrument platforms would benefit from the implementation of GPS and compass instrumentation. These digital measurements would enable real time processing of cardinal wind direction and global position. Currently, the algorithm tha t aligns the disdrometers to the mean rain vector assumes a mean drop diameter of 1.2 mm, a valid assumption for extreme rainfall scenarios. A new tracking algorithm that aligns the disdrometer to the measured mean drop size may expand the use of the inst rumentation platforms to other applications and decrease the occurrence of, otherwise unknown, incorrectly measured data.

PAGE 133

133 PARSIVEL validation results from this research project indicate that the error introduced by oblique trajectory angles manifest as inc orrect velocities, while droplet diameters are measured correctly. A new data processing algorithm may be created to utilize real time wind data for the calculation of droplet velocities rather than residence time in the measurement field. This new measu rement technique may provide a way of increasing the maximum acceptable wind speed when using stationary instrumentation. Recommendations for Full Scale and Numerical Models The implementation of a full scale WDR simulation system that accurately simulate s naturally occurring RSDs may be used in the validation of numerical WDR models by implementing similar structures in the models and the IBHS Research Facility. The full scale simulation system will also prove useful in the evaluation of building componen t performance. The data gathered from field measurements can be used in time varying full scale simulations of extreme WDR scenarios. Thus, future experiments could compare the performance of building components subjected to current testing protocols and simulations based on field measured data to establish the efficacy current standardized test methods. determine the driving forces of particles moving through space. Implementing a wi nd field determined by a CFD software package to the developed trajectory model will better results and warrants further research.

PAGE 134

134 Recommendations for the Morphological Image Proc essing Algorithm The design of a robust graphical user interface for the morphological image processing algorithm would serve as an economical solution to RSD measurement. When compared to disdrometric instrumentation, it is a labor intensive process; howe ver, it may be a useful tool for applications where the drop size distribution is known to remain constant or for instrumentation validation. trajectory model are made available t o the public, they may be used by nozzle manufacturers and designers for the characterization of nozzles and for the design of currently available.

PAGE 135

135 APPENDIX Theoretic al Proof of Greater Accuracy f rom an Articulating Instrument This proof follows the work of Griffiths (1974) in which he demonstrated that the sample volume for an articulating instrument is different than that of a st ationary instrument Ultimately, it is shown that the articulating sensor has a greater accuracy due to the greater sample volume. Case 1 represents a stationary instrument oriented such that the sample area is oriented parallel to the ground, Case 2 and 3 represent an articulating instrument platform such that the sample area is tilted to angle (determined by the trajectory of the assumed mean droplet diameter) and other droplets are crossing the sample area at angle depending whether the droplet is smaller than assumed mean (Case 2) or larger (Case 3). S = Sample ar ea (m 2 ) V T D = Terminal velocity of droplets of diameter D (m/s) U = Longitudinal wind velocity (m/s) V R D = Resultant velocity of droplets of diameter D (m/s) C D = Number density of droplets of diameter D (m 1 ) n D = Number of droplets of diameter D ha 1 ) V S = Sample volume (m 3 ) M D N D = Number concentration of droplets of diameter D Case 3 Case 2 Case 1

PAGE 136

136 Stationary instrument (Case 1): If we collect data for Time T and say total number of droplets = M : We know: Thus: Solve for concentration of droplets: Giv en: Thus we find that the sample volume for a stationary instrument is:

PAGE 137

137 Articulating instrumentation (Case 2 & 3) Using the following identities: We get the following for both Case 2 and 3: We know: Substitution gives us: Checks Say : If we collect data for Time T and say total number of droplets = M : Solve for concentration of droplets: Thus we find that the sample volume for an articulating instrument is:

PAGE 138

138 C ompare Instrumentation platforms: Given we know: for stationary instrumentation for articulating instrumentation K can vary from U to V T,D If the tracking algorithm is acc urate, in high winds K approaches U and V T,D in no wind for stationary instrumentation for articulating instrumentation in no wind for articulating instrumentation in high winds Therefore as the wind speed increases the sample volume of the articulating instrument platform increases, making it more accurate. Example: Assuming that the mean drop diameter is 1.2 mm and U = 10 m/s: for stationary instrumentation for articulating instrumentation Thus the sample volume of articulating instrument is gr eater than twice as large at U = 10 m/s. Similarly the ratio of the sample volumes can be calculated for multiple wind velocities:

PAGE 139

139 Nozzle Selection Nozzles are used in a wide range of applications, including evaporative cooling, gas conditioning, fir e suppression, spray drying, and agriculture. Nozzles generate droplets through the process of atomization, in which a fluid with a potential energy (water pressure) is released through an opening (nozzle end) and emerge as ligaments that evolve into drop lets (Schick 1997). The resultant drop size distribution is a function of the nozzle type, spray type, spray angle, and the pressure and flow rate of the fluid. Figure A 1 Droplet formation process

PAGE 140

140 Figure A 2 Types of nozzles and drop size relati onship The effect of each variable is as follows: Nozzle types which can be hydraulic or air assisted. A ir assisted nozzles generally produce droplets in a smaller size range than hydraulic nozzles Spray types are flat, solid stream, square, hollow cone and full cone sprays. Full cone and square spray nozzles produce the largest drop size distribution followed by flat spray and hollow cone spray nozzles (Fig 4 2). Spray angle the angle through which the spray is effective is inversely proportional to d rop size. Nozzle pressure is inversely proportional to drop size and is controlled by the pumping system. Flow rate is directly proportional to the drop size and controlled by an actively controlled valve system.

PAGE 141

141 Measured Diameter Wind Relationships H urricane Ike Data

PAGE 142

142

PAGE 143

143

PAGE 144

144 Hurricane Irene (Beaumont) Data

PAGE 145

145

PAGE 146

146

PAGE 147

147 Hurricane Irene (Deal) Data

PAGE 148

148

PAGE 149

149

PAGE 150

150 F lorida C oastal M onitoring P rogram Hurricane Ike Data

PAGE 151

151

PAGE 152

152

PAGE 153

153 Verification of the Origins of Rotation in Tornadoes Experiment 2 Articulating Instrume nt Platform Data

PAGE 154

154

PAGE 155

155

PAGE 156

156

PAGE 157

157

PAGE 158

158

PAGE 159

159

PAGE 160

160

PAGE 161

161

PAGE 162

162

PAGE 163

163

PAGE 164

164

PAGE 165

165

PAGE 166

166

PAGE 167

167

PAGE 168

168

PAGE 169

169

PAGE 170

170

PAGE 171

171

PAGE 172

172

PAGE 173

173

PAGE 174

174

PAGE 175

175

PAGE 176

176

PAGE 177

177

PAGE 178

178

PAGE 179

179

PAGE 180

180

PAGE 181

181

PAGE 182

182

PAGE 183

183

PAGE 184

184

PAGE 185

185

PAGE 186

186

PAGE 187

187

PAGE 188

188

PAGE 189

189

PAGE 190

190

PAGE 191

191

PAGE 192

192

PAGE 193

193

PAGE 194

194

PAGE 195

195

PAGE 196

196

PAGE 197

197

PAGE 198

198

PAGE 199

199

PAGE 200

200

PAGE 201

201

PAGE 202

202

PAGE 203

203

PAGE 204

204

PAGE 205

205

PAGE 206

206

PAGE 207

207

PAGE 208

208

PAGE 209

209

PAGE 210

210 Measured Nozzle Characteristics Uniformity BETE WL 1

PAGE 211

211 BETE WL 2

PAGE 212

212 BETE WL 1 1/2

PAGE 213

213 BETE WL 3

PAGE 214

214 Lechler 608

PAGE 215

215 Lechler 648

PAGE 216

216 Spray Systems 1/8GG

PAGE 217

217 Steinen 100

PAGE 218

218 Steinen 150

PAGE 219

219 Measured Nozzle R aindrop S ize D istribution

PAGE 220

220

PAGE 221

221

PAGE 222

222

PAGE 223

223

PAGE 224

224 REFERENCES AAMA 2005. AAMA 520. Voluntary specification for rating the severe wind driven rain resistance of windows doors and unit skylights. Illinois. Abuku, M, H Janssen, and S Roels, 2009. Impact of wind driven rain on historic brick wall buildings in a moderately cold and humid climate: Numerical analyses of mould growth risk, indoor climate and energy consumption. Energy and building 41:101 110. Adams, R, 1 993. Radial Decomposition of Discs and Spheres. Computer Vision, Graphics, and Image Processing: Graphical Models and Image Processing 55(5):325 332. ASTM 2000. ASTM E331 00. Standard Test Method for Water Penetration of Exterior Windows, Skylights, Doors, and Curtain Walls by Uniform Static Air Pressure Difference. Pennsylvania. ASTM 2000. ASTM E547 00. Standard Test Method for Water Penetration of Exterior Windows, Skylights, Doors, and Curtain Walls by Cyclic Static Air Pressure Difference. Pennsylvania. ASTM 2004. ASTM E2268 04. Standard Test Method for Water Penetration of Exterior Windows, Skylights and Doors by Rapid Pulsed Air Pressure Difference. Pennsylvania. ASTM 2005. ASTM E1105 05. Standard Test Method for Field Determination of Water Penetratio n of Installed Exterior Windows, Skylights, Doors and Curtain Walls Uniform or Cyclic Static Air Pressure. Pennsylvania. Atlas, D, CW Ulbrich, FD Marks, E Amitai, and CR Williams, 1999. Systematic variation of drop size distribution and radar rainfall rela tion. J. Geophys. Res. 104:6155 6169. Balderrama, JA, FJ Masters, KR Gurley, DO Prevatt, LD Aponte Bermdez, TA Reinhold, JP Pinelli, CS Subramanian, SD Schiff, and AG Chowdhury, 2011. The Florida Coastal Monitoring Program (FCMP): A review. Journal of Win d Engineering and Industrial Aerodynamics 99(9):979 995. Beckett, HE 1938. Building Research Note No. 755. Bendat, JS and AG Piersol, 2000. Random Data Analysis and Measurement Procedures (third edition) John Wiley & Sons. Best, AC, 1950. The Size Distr ibution of Raindrops. Quarterly Journal Royal Meteorological Society 76(327):16 36.

PAGE 225

225 Bitsuamlak, GT, AG Chowdhurry, and D Sambare, 2009. Application of a Full Scale Testing Facility for Assessing Wind Driven Rain Intrusion. Building and Environment 44:2430 2441. Blackall, TN and MC Baker 1984. Rain Leakage of Residential Windows in the Lower Mainland of British Columbia. Blocken, B and J Carmeliet, 2002. Spatial and Temporal Distribution of Driving Rain On A Low Rise Building. Wind and Structures 5(5):441 46 2. Blocken, B and J Carmeliet, 2004. A Review of Wind Driven Rain Research in Building Science. Journal of Wind Engineering and Industrial Aerodynamics 92:1079 1130. Blocken, B and J Carmeliet, 2005. Guidelines for Wind, Rain and Wind Driven Rain Measureme nts at Test Building Sites. The 7th symposium on Building Physics in the Nordic Countries Reykjavk, Iceland Blocken, B and J Carmeliet, 2010. Overview of Three State of The Art Wind Driven Rain Assessment Models and Comparison Based on Model Theory. Buil ding and Environment 45:691 703. Blocken, B, G Dezso, J van Beek, and J Carmeliet, 2010. Comparison of Calculation Models for Wind Driven Rain Deposition on Building Faades. Atmospheric Environment 44:1714 1725. Bomberg, MT, MZ Rousseau, G Desmarais, M Ni cholls, and MA Lacasse, 2002. Report from Task 2 of MEWS Project Description of 17 Large Scale Wall Specimens Built for Water Entry Investigation in IRC Dynamic Wall Testing Facility. Bondi, P and Stefanizzi, 2001. Hygro thermal performance of hollow brick s and current standards. Energy and Buildings 33:731 736. Bottema, M 1993. Wind climate and urban geometry. Ph.D. Thesis, Eindhoven. Bradley, SG and CD Stow, 1974. The Measurement of Charge and Size of Raindrops: Part 1. The Disdrometer. Journal of Applied Meteorology 13:114 130. Bradley, SG and CD Stow, 1975. Reply. Journal of Applied Meteorology 14:426 428. Cao, Q and G Zhang, 2009. Errors in Estimating Raindrop Size Distribution Parameters Employing Disdrometer and Simulated Raindrop Spectra. Journal of Applied Meteorology and Climatology 48:406 425. Chang, WY, TCC Wang, and PL Lin, 2009. Characteristics of the Raindrop Size Distribution and Drop Shape Relation in Typhoon Systems in the WEsten Pacific from the 2D Video Disdrometer and NCU C Band Polar imetric Radar. Journal of Atmospheric and Oceanic Technology 1973 1993.

PAGE 226

2 26 Chen, JP and D Lamb, 1994. Simulation of cloud microphysical and chemical processes using a multicomponent framework. Part I: Description of the microphysical model. Journal of Atmos pheric Science 51:2613 2630. Choi, ECC, 1993. Simulation of Wind Driven Rain Around the Building. Journal of Wind Engineering and Industrial Aerodynamics 46 47:721 729. Choi, ECC, 1994. Determination of Wind Driven Rain Intensity on Building Faces. Journal of Wind Engineering and Industrial Aerodynamics 51(1):55 69. Choi, ECC, 1994. Parameters Affecting the Intensity of Wind Driven Rain on the Front Face of a Building. Journal of Wind Engineering and Industrial Aerodynamics 53(1 2):1 17. Choi, ECC, 1996. Me asurement of Wind Driven Rain. Proceedings of the fifth Australian Wind Engineering Society Workshop ,(pp. A 6). Adelaide Choi, ECC, 1999. Wind Driven Rain on Building Faces and the Driving Rain Index. Journal of Wind Engineering and Industrial Aerodynamics 79(1 2):105 122. Clift, R and WH Gauvin, 1970. The motion of particles in turbulent gas streams. Proc. Chem. E.C.A. 1:14 24. Cornick, S and WA Dalgliesh, 2009. Adapting rain data for hygrothermal models. Building and Environment 44(5):987 996. Counihan, J 1975. Adiabatic Atmospheric Boundary Layers Review and Analysis of Data From Period 1880 1972. Atmospheric Environment 9(10):871 905. Dao, TN and JW Van de Lindt, 2010. Methodology for Wind Driven Rainwater Intrusion Fragilities for Light Frame Wood Roof Systems. Journal of Structural Engineering 136(6):700 706. Davenport, AG, 2002. Past, present and future of wind engineering. Journal of Wind Engineering and Industrial Aerodynamics 90:1371 1380. Derome, D and P Fazio, 2000. Large scale testing of two fla t roof assemblies insulated with cellulose. Journal of Architectural Engineering (American Society of Civil Engineers) 6(1):12 13. Donnadieu, G, 1980. Comparison of. Comparison of Results Obtained With the VIDIAZ Spectropluviometer and the Joss Waldvogel Rainfall Disdrometer in a "Rain of Thunder Type" 19:593 597. Fangmeir, DD, WJ Elliot, SR Workman, RL Huffman, and GO Schwab, 2006. Soil and Water Conservation Engineering 5th Ed New York: Thomson Delmar Learning.

PAGE 227

227 Federal Emergency Management Agency 2005 FEMA 490, Summary Report on Building Performance: 2004 Hurricane Season. Washington DC. Federal Emergency Management Agency 2006. FEMA 549, Summary Report on Building Performance: Hurricane Katrina 2005. Washington DC. Flower, JW and TV Lawson, 1972. On the Laboratory Representation of Rain Impingement on Buildings. Atmospheric Environment 6:55 60. Friedrich, K, S Higgins, F Masters, and C Lopez, 2011. Artifacts in PARSIVEL disdrometer measurements under strong wind conditions. Journal of Atmospheric and Oceanic Technology In Press. Gaudet, B and WR Cotton, 1998. Statistical characteristics of a real time precipitation forecasting model. Weather Forecasting 13:966 982. Gray, W M., 1979. Hurricanes: Their formation, structure and likely role in the tropical circulation. Meteorology Over Tropical Oceans.D.B. Shaw (Ed.), Royal Meteorological Society 155 218. Griffiths, RF, 1974. Comments on "The Measurement of Charge and Size of Raindrops: Parts I and II". Journal of Applied Meteorology 14:422 425. Grossklau s, M, K Uhlig, and L Hasse, 1998. An Optical Disdrometer for Use in High Wind Speeds. Journal of Atmospheric and Oceanic Technology 15:1051 1059. Guillermo, F, R Green, B Khazai, A Smyth, and G Deodatis, 2010. Field Damage Survey of New Orleans Homes in the Aftermath of Hurricane Katrina. Natural Hazards Review 11:7 18. Gunn, R and GD Kinzer, 1949. The terminal velocity of fall for water droplets in stagnant air. Journal of Meteorology 6:243 248. Gurley, K, J Burton, M Abdullah, F Masters, and T Reinhold 2006. Post 2004 Hurricane Field Surveyan Evaluation of the Relative Performance of the Standard Building Code and the Florida Building Code. Structures Research Communication No. 53102 2, Gainesville, FL. Gurley, K and FJ Masters, 2011. Post 2004 Hurricane Field Survey of Residential Building Performance. Natural Hazards Review Accepted for publication. Habib, E, B Larson, and J Graschel, 2009. Validation of NEXRAD Multisensor Precipitation Estimates Using an Experimental Denses Rain Gauge Network in South Louisiana. Journal of Hydrology 373:463 478. Hartley, AJ, G Chong, J Houston, and E Mather, 2005. 150 Million years of climatic stability: evidence from the Atacama Desert, Northern Chile. Journal of the Geological Society 162:421 424.

PAGE 228

228 Hauser, D, P Amayenc B Nutten, and P Waldteufel, 1984. A New Optical Instrument for Simultaneous Measurements of Raindrop Diameter and Fall Distribution. Journal of Atmospheric and Oceanic Technology 1:256 269. Hoppestad, S 1955. Slagregn i Norge. Norwegian Building Resear ch Institute, Oslo. Houze, R A., 1994. Cloud Dynamics Academic Press. Hudson, NW, 1970. Raindrop Size Distribution in High Intensity Storm. Rhodesian Journal of Agricultural Research 1:6 11. Hulya, K, K Nygren, and P Norberg, 2004. In use performance asse ssment of rendered autoclaved aerated concrete walls by long term moisture monitoring. Building and environment 39:677 687. Hu, Z and RC Srivastava, 1995. Evolution of raindrop size distribution by coalescence, breakup, and evaporation: Theory and Observat ion. Journal of Atmospheric Sciences 52(10):1761 1783. Illingworth, AJ and CJ Stevens, 1987. An Optical Disdrometer for the Measurement of Raindrop Size Spectra in Windy Conditions. Journal of Atmospheric and Oceanic Technology 4:411 421. Inculet, D and D Surry 1994. Simulation of Wind Driven Rain and Wetting Patterns on Buildings. Boundary Layer Wind Tunnel Report : BLWTL SS30 1994. Insurance Information Institute. (2011). www.iii.org. Retrieved August 31, 2011 (www.iii.org/disasters/Hurricane). Insuranc vs. Structural Strength. www.disastersafety.org/resource/resmgr/pdfs/hurricane_ike.pdf. ISO, BS E. 2009. Hygrothermal performance of buildings calculation and presentation of clim atic data Part 3: calculation of a driving rain index for vertical surfaces from hourly wind and rain data. BS EN ISO 15927 3:2009 Jamison, AR and AB Kostinski, 2000. Fluctuation properties of precipitation. Part VI: Observations of hyperfine clusterin g and drop size distribution structures in three dimensional rain. Journal of the Atmospheric Sciences 57(3):373 388. Jones, R and P Soille, 1996. Periodic lines: Definition, cascades, and application to granulometrie. Pattern Recognition Letters 17:1057 1 063. Joss, J and A Waldvogel, 1967. A raindrop spectrometer with automatic readout. Pure Applied Geophysics 68:240 246.

PAGE 229

229 Joss, J and A Waldvogel, 1970. A method to improve the accuracy of radar measured amounts of precipitation. Preprints, 14th Conf. on Rad ar Meteorology ,(pp. 237 238). Tucson, AZ Karagiozis, A, G Hadjisophocleous, and S Cao, 1997. Wind Driven Rain Distributions on Two Buildings. Journal of Wind Engineering and Industrial Aerodynamics 67 & 68:559 572. Kedem, B, H Pavlopoulos, X Guan, and DA S hort, 1994. A probability distribution model for rain rate. Journal of Applied Meteorology 33:1486 1493. MEWS Project: Experimental assessment of water penetration and entry into wood frame wall specimens Final report. Ottowa, Ont. Lacy, RE, 1951. Observations With a Directional Raingauge. Quarterly Journal of The Royal Meteorological Society 77(332):283 292. Lacy, RE, 1965. Driving rain maps and the onslaught of rain on buildings RILEM/CIB Symposium on Moisture Problems in Buildings, Rain Penetration ,(pp. 3 4). Helsinki Lang, A, M Lawton, and WC Brown 1999. Stucco clad wall drying experiment. Research report no. 5972204.00, Vancouver, B.C. Lim, HC, TG Thomas, and IP Castro, 2009. Flow around a cube in a turbulent boundary layer: LES and experiment. Journal of Wind Engineering and Industrial Aerodynamics 97(2):96 109. Linsley, RK, JB Franzini, DL Freyberg, and G Tochbanoglous, 1992. Water resources engineering, fourth ed New York: Irwin McGraw Hill. Loffler Mang, M and J Joss, 2000. An Optical Disdrometer for Measuring Size and Velocity of Hydrometeors. Journal of Atmospheric and Oceanic Technology 17:130 139. Lonfat, M, FD Marks, and SS Chen, 2004. Precipitation Distribution in Tr opical Cyclones Using the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager: A Global Perspective. Monthly Weather Review 132(7):1645 1660. Lopez, CR 2009. Comparison of wind driven rain test methods for fenestration. ME Thesis, Gainesville, Fl. Lopez, CR, 2011. Water Penetration Resistance of Residential Window and Wall Systems Subjected to Steady and Unsteady Wind Loading. Building and Environment 46:1329 1342.

PAGE 230

230 Lopez, CR, FJ Masters, and SR Bolton, 2011. Water Penetration Resistance of Residen tial Window and Wall Systems Subjected to Steady and Unsteady Wind Loading. Building and Environment 46:1329 1342. Lstiburek, JW 2004. Rainwater Management Performance of Newly Constructed Residential Building Enclosures During August and September 2004. Report Prepared for the Home Builders Association of Metro Orlando and the Florida Home Builders Association. Lstiburek, JW 2005. Rainwater Management Performance of Newly Constructed Residential Building Enclosures During August and September 2004. Westf ord, MA. Marshall, JS, RC Langille, and WM Palmer, 1947. Measurement of Rainfall by Radar. Journal of Meteorology 4:186 192. Marshall, JS and WM Palmer, 1948. The Distribution of Raindrops With Size. Journal of Meteorology 5:165 166. Masters, FJ, KR Gurley and DO Prevatt, 2008. Full scale simulation of turbulent wind driven rain effects on fenestration and wall systems. Full scale simulation of turbulent wind driven rain effects on fenestration and wall systems Tokyo Masters, FJ, HW Tieleman, and JA Balde rrama, 2010. Surface Wind Measurements in Three Gulf Coast Hurricanes of 2005. Journal of Wind Engineering and Industrial Aerodynamics 98:533 547. Masters, FJ, P Vickery, P Bacon, and E Rappaport, 2010. Towards Objective Standardized Intensity Estimates fr om Surface Wind Speed Observations. Bulletin of the American Meteorological Society 1665 1681. Medlin, JM, SK Kimball, and KG Blackwell, 2007. Radar and Rain Gauge Analysis of Monthly Weather R eview 135:1869 1888. Mehta, KC, JE Minor, and TA Reinhold, 1983. Wind Speed Damage Correlation in Hurricane Frederic. Journal of Structural Engineering 1:37 49. Michaelides, EE, 2006. Particles, Bubbles & Drops Hackensack, NJ: World Scientific Publishing Co. Pte. Ltd. Mualem, Y and S Assouline, 1986. Mathematical Model for rain Drop Distrubution an Rainfall Kinetic Energy. Transactions of the American Society of Agricultural Engineering ASAE TAAEAJ 29(2):494 500. Mullens, M, B Hoekstra, I Nahmens, and F Ma rtinez 2006. Water Intrusion in Central Florida Homes During Hurricane Jeanne in September 2004. Orlando, FL.

PAGE 231

231 Mullens, M, B Hoekstra, I Nahmens, and F Martinez 2006. Water Intrusion in Central Florida Homes during Hurricane Jeanne in September 2004. Report to U.S. DOE, Orlando, Fl. National Institue of Standards and Technology 2005. NIST Technical Note 1476 Performance of Physical Structures in Hurricane Katrina and Hurricane Rita: A Reconnaissance Report. Nespor, V, WF Krajewski, and A Kruger, 2000. Wind Induced Error of Raindrop Size Distribution Measurement Using a Two Dimensional Video Disdrometer. Journal of Atmospheric and Oceanic Technology 17:1483 1492. Nespor, V, WF Krajewski, and A Kruger, 2000. Wind Induced Error of Raindrop Size Distribution M easurement Using a Two Dimensional Video Disdrometer. Journal of Atmospheric and Oceanic Technology 17:1483 1492. Newland, DE, 1993. An Introduction to Random Vibrations, Spectral & Wavelet Analysis Dover Publications Inc. NOAA 1977. NOAA Technical Memo randum NWS Hydro 35. Technical Memorandum. Nore, K, B Blocken, BP Jelle, JV Thue, and J Carmeliet, 2007. A dataset of wind driven rain measurements on a low rise test building in Norway. Building and Environment 42:2150 2165. NWS 1961. Technical Paper No. 40 Rainfall Frequency Atlas of The United States for Durations from 30 Minutes to 24 Hours and Return Periods from 1 to 100 Years. Panton, R, 2005. Incompressible Flow (third edition) John Wiley & Sons. Poss, DB 2000. Design and evaluation of a mobile w ind instrumentation tower for hurricane wind measurements. Masters Thesis, Civil Engineering, Clemson University, Clemson, SC. Rasmussen, EN, J Straka, R Davies Jones, CA Doswell, FH Carr, MD Eilts, and DR MacGorman, 1994. Verification of the Origins of Ro tation in Tornadoes Experiment: VORTEX. Bulletin of the American Meteorological Society 75(6):995 1006. Rayment, R and M Hilton, 1977. The Use of Bubbles in a Wind Tunnel for Flow Visualisation and the Possible Representation of Raindrops. Journal of Indus trial Aerodynamics 2:149 157. RDH 2002. Water Penetration Resistance of Windows Study of Codes, Standards, Testing, and Certification. RDH 2002. Water Penetration Resistance of Windows Study of Manufacturing, Building Design, Installation and Maintenance Factors.

PAGE 232

232 Richter, C and M Hagen, 1997. Drop Size Distributions of Raindrops by Polarization radar and simultaneous measurements with disdrometer, wind profiler, and PMS Probes. Quarterly Journal of the Royal Meteorological Society 123(544):2277 2296. Rodg ers, GG, G, P. J. Poots, and WM Pickering, 1974. Theoretical Predictions of Raindrop Impaction on a Slab Type Building. Building Science 9:181 190. Rosenfeld, D, DB Wolff, and D Atlas, 1993. General Probability matched Relations between Radar Reflectivity and Rain Rate. Journal of Applied Meteorology 32:50 72. Salzano, C, F Masters, and J Katsaros, 2010. Water Penetration Resistance of Residential Window Installation Options for Hurricane Prone Areas. Building and Environment 45(6):1373 1388. Sauvageot, H a nd JP Lacaux, 1994. The shape of averaged drop size distributions. Journal of Atmospheric Sciences 52(8):1070 1083. Schafer, R, S Avery, P May, D Rajopadhyaya, and C Williams, 2002. Estimation of Rainfall Drop Size Distribution from Dual Frequency Wind Pro filer Spectra using Deconvolution and a Nonlinear Least Squares Fitting Technique. Journal of Atmospheric and Oceanic Technology 19:964 874. Schiller, L and A Nauman, 1933. Uber die grundlegende Berechnung bei der Shwekraftaufbereitung. ver. Deutch Ing. 44:318 320. Sheppard, BE, 1990. Effect of Irregularities in the Diameter Classification of Raindrops by the Joss Waldvogel Disdrometer. Journal of Atmospheric and Oceanic Technology 7: 180 183. Simpson, RH, 1974. The Hurricane Disaster Potential Scale. Weatherwise 27:169 186. Simpson, RH and H Riehl 1981. The Hurricane and its Impact. Smith, JA, ML Baeck, Y Zhang, and CA Doswell, 2000. Extreme Rainfall and Flooding from Supercell Thunder storms. Journal of Hydrometeorology 2(5):469 489. Stoelinga, MT, PV Hobbs, CF Mass, JD Locatelli, BA Colle, RA Houze, AL Rangno, NA Bond, BF Smull, RM Rasmussen, G Thompson, and BR Colman, 2003. Improvement of Microphysical Parameterization through Observa tional Verification Experiment. Bulletin of the American Meteorological Society 84:1807 1826. Straube, JF and EFP Burnett, 2000. Simplified Prediction of Driving Rain Deposition. Proceedings of International Building Physics Conference ,(pp. 375 382). Eindh oven

PAGE 233

233 Straube, J, C Schumacher, and E Burnett 1995. In Service Performance of Enclosure Walls: Construction and Instrumentation report, Report N2L 3G1 BEG Building Engineering Group. Ontario. Surry, D, DR Inculet, PF Skerlj, JX Lin, and AG Davenport, 1994. Wind Rain and the Building Envelope: a Status Report of Ongoing Research at the University of Western Ontario. Journal of Wind Engineering and Industrial Aerodynamics 53:19 36. Tattelman, P and KG Scharr, 1983. A Model for Estimating One Minute Rainfall Ra tes. Journal of Climate and Applied Meteorology 22:1575 1580. Teasdale St Hilaire, A and D Derome, 2007. Comparison of Experimental and Numerical Results of Wood Frame Wall Assemblies Wetted by Simulated Wind Driven Rain Infiltration. Building and Environm ent 39:1131 1139. Testud, J, S Oury, RA Black, P Amayenc, and X Dou, 2001. The Concept of "Normalized" Distribution to Describe Raindrop Spectra: A Tool for Cloud Physics and Cloud Remote Sensing. Journal of Applied Meteorology 40:1118 1140. Tokay, A, PG Bashor, E Habib, and T Kasparis, 2008. Raindrop Size Distribution Measurements in Tropical Cyclones. American Meteorological Society Monthly Weather Review 136(5):1669 1684. Tokay, A, MD Greenbelt, KR Wolff, P Bashor, and OK Dursun, 2003. On the Measurem ent Errors of the Joss Waldvogel Disdrometer. 31st International Conference on Radar Meteorology Tokay, A, A Kruger, and WF Krajewski, 2001. Comparison of Drop Size Distribution Measurements by Impact and Optical Disdrometers. Journal of Applied Meteorolo gy 40:2083 2097. Tokay, A and DA Short, 1996. Evidence from tropical raindrop spectra of the origin of rain from stratiform versus convective clouds. Journal of Applied Meteorology 35:355 371. Tokay, A, DA Short, RW Christopher, WL Ecklund, and K Gage, 1 999. Tropical rainfall associated with convective and stratiform clouds: Intercomparsion of disdrometer and profiler measurements. Journal of Applied Meteorology 38:302 320. Torres, DS, J Porra, and D Creutin, 1994. A General Formulation for Raindrop Size Distribution. Journal of Applied Meteorology 33:1494 1502. Tsongas, GA, DP Govan, and JA McGillis, 1998. Field observations and laboratory tests of water migration in walls with shiplap hardboard siding. American Society of Heating, Refrigerating and Air C onditioning Engineers Thermal Performance of the Exterior Envelopes of Buildings 7:469 483.

PAGE 234

234 Ulbrich, CW, 1983. Natural Variations in the Analytical Form of the Raindrop Size Distribution. Journal of Climate and Applied Meteorology 22:1764 1775. Underwood, SJ and V Meentemeyer, 1998. Climatology of Wind Driven Rain for the Contiguous United States for the Period 1971 to 1995. Physical Geography 19(6):445 462. van den Boomgaard, R and Richard van Balen, 1992. Methods for fast morphological image transforms us ing bitmapped binary images. CVGIP: Graphical Models and Image Processing 54(3):252 258. Weber, RO, 1999. Remarks on the definition and estimation of friction velocity. Boundary Layer Meteorology 93:197 209. Willis, PT and P Tattelman, 1989. Drop Size Distribution Associated With Intense Rainfall. Journal of Applied Meteorology 28:3 15. Wilson, J W. and E Brandes, 1979. Radar Measurement of Rainfall A Summary. Bulletin American Meteorological Society 60(9):10 48 1058. Zhang, G, J Vivekanandan, and EA Brandes, 2003. The Shape Slope Relation in Observed Gamma Raindrop Size Distributions: Statistical Error or Useful Information? Journal of Atmopheric and Oceanic Technology 20:1106 1119.

PAGE 235

235 BIOGRAPHICAL SKETCH Ca rlos Rodolfo Lopez was born in Bogota, Colombia At the age of five, he and his parents moved to the state of Florida where he was raised. As the only child of a b usiness a dministrator and a rchitect/ c ontractor he was exposed to the construction environme nt at a very young age. Upon graduating from John I. Leonard High School in June 2003 he began his pursuit of a Bachelor of Science degree in civil engineering, with a focus in Structures, from the University of Florida. While completing his B.S. he had the privilege of meeting great professionals from his field who shared his interests and passions. Upon completion of his degree in 2007 he sought the mentorship from one such professional, Dr. Forrest J. Masters. Under his mentorship he studied the behavi or of building assemblies subjected to hurricane force wind and rain. In his study of building science he had the great opportunity of working side by side with professionals ranging from top executives of some the largest fenestration manufacturing compan ies to heads of the top trade organizations and leading legislative officials. Carlos also had the privilege participating in the Florida Coastal Monitoring Program (FCMP). The FCMP is a unique joint venture focusing on full scale experimental methods th at quantify low level hurricane wind behavior and the resultant loads on residential structures As a team leader of the FCMP Carlos deployed to Hurricanes Gustav in Louisiana, Ike in Texas, Irene in North Carolina, and Tropical Storm Fay in Central Flori da, to set up instrumentation that quantified near surface hurricane behavior. Upon passing of the storms, Carlos also participated in teams that performed post storm damage assessments. In the fall of 2011, he received his Ph.D. from the University of Fl orida.

PAGE 236

236 Carlos R. Lopez is a student member of the American Association for Wind Engineering, the American Society of Civil Engineers, and American Concrete Institute