To my parents, Amanda Duarte and Alfonso...

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

2

© 2011 Carlos R. Lopez

3

To my parents, Amanda Duarte and Alfonso Lopez

4

ACKNOWLEDGMENTS

I would like to thank my advisor, Forrest J. Masters, Ph.D., P.E. and committee

members Kurtis R. Gurley, 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

graduate career. I would also like to extend my appreciation to George Fernandez,

Jason Smith, James Austin, Juan 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.

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 Façade ........................... 31 Factors Affecting Rain Deposition Rate on the Building Façade ....................... 32

Rainfall intensity ............................................................................................ 32

Influence of the wind on raindrop trajectory ................................................. 33 Terrain characteristics ................................................................................... 36 Building characteristics ................................................................................. 37

Modeling of Rain Deposition on the Building Façade ............................................... 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

6

Instrumentation Used In This Study .................................................................... 47 OTT PARSIVEL optical disdrometer ............................................................ 47

Droplet 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 Distribution 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 Distribution 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

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 Rain 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 Comparison 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

8

Uniformity ........................................................................................................... 210 Measured Nozzle Raindrop Size Distribution ................................................... 219

REFERENCES ................................................................................................................ 224

BIOGRAPHICAL SKETCH.............................................................................................. 235

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

10

LIST OF FIGURES

Figure page 1-1 Articulating precipitation measurement 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 trajectory 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

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 system and Gill anemometers at ............ 72

3-15 Measured raindrop size distribution by stationary instrumentations and articulating instrumentation .................................................................................... 74

3-16 Comparison 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 diameter 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 Observed shape slope relation .............................................................................. 94

12

4-13 Model raindrop size distribution and measured raindrop size distribution comparisons ........................................................................................................... 96



4-16 Mean square error values of raindrop size distribution models stratified by U and R ...................................................................................................................... 99

4-17 Mean square error values of raindrop size distribution models stratified by TIU and R ...................................................................................................................... 99

4-18 Mean square error values of raindrop size distribution models stratified by TIV 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 distributions 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 footage ......................................... 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 heights and wind velocities .... 124

5-9 Comparison of measured raindrop size distributions and Best (1950) model ... 125

13

A-1 Droplet formation process .................................................................................... 139

A-2 Types of nozzles and drop size relationship ....................................................... 140

14

Abstract of Dissertation Presented to the Graduate School of the University 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 Engineering

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

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 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.

15

CHAPTER 1

INTRODUCTION

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

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 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 this 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.

The motivation for this research is the extensive damage caused by tropical

cyclones annually. In 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 construction is water ingress through the

building envelope. Wind 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). WDR deposition on the building

16

façade 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 science,

engineering, and building science disciplines. Engineering and building science

research has primarily focused on modeling wind-driven rain deposition on building

façades (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 improvement 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).

Raindrop size distribution (RSD) is a critical variable in both fields. One view of

meteorology considers the RSD as a microphysical ‘signature’ of much larger scale

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

17

weather systems are particularly sensitive to the drop size. In engineering, the wetting

pattern and the rain deposition rate on the building façade is, in part, a function of the

RSD.

Since the 1940s, a significant amount of RSD data has been collected during

stratiform and convective 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 façade. This

research directly addresses 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 science) 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 winds. 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). Data obtained during the VORTEX2 campaign 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 from the results of

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 continuously

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 raindrop passing through a light band (Loffler-Mang and Joss, 2000).

Reliable rain data acquisition via optical 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 stationary 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 actively 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

19

were implemented. The first 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 with 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 mechanized turret on a Florida Coastal Monitoring Program (FCMP) weather station (Figure 1-2). This instrument was first deployed successfully during

Hurricane Ike (2008).

Figure 1-1. Articulating precipitation measurement platform (left) and stationary platform

(right, photo courtesy of author)

20

Figure 1-2. Portable FCMP weather station with the actively controlled positioning system (photo courtesy of author)

Thrust 2: Characterization of the Raindrop Size Distribution in Atlantic Tropical Cyclones and Supercell Thunderstorms in the Midwest U.S.

With the use of the new instrument platforms, multiple measurements were taken

in supercell thunderstorms and in tropical cyclones as part of the VORTEX2 and FCMP

field campaigns. These data, to the author’s knowledge, are the first set of RSD

measurements in high winds. Thus, the analysis and results of these data are presented

in this document to address the following questions in an effort to advance the WDR

knowledge base:

1. Does 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 little-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?

21

4. Does the precipitation algorithm in 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 simulation and specification of design requirements for water penetration resistance of building products and systems?

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

In addition to answering the preceding questions, the data was also used to

develop a method 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 research 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 determine the optimal choice for the

facility.

Figure 1-3. Insurance Institute for Business & Home Safety Research Center (photo courtesy of IBHS)

IBHS.ORG

22

Importance of the Study

The research findings have the potential 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 array. 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 computational wind engineering research, to model WDR trajectories, current

analysis techniques employ the following aspects: (1) the wind flow pattern is calculated

by solving the three dimensional Reynolds Averaged Navier-Stokes equations, the

continuity equation, and the equations of the realizable k-ε 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 Palmer 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

23

simulating extreme WDR conditions. This information will be made available to the

American Society of Civil Engineers’ (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 only 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 façade

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 system, 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 overview 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 evaluation 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

24

6 summarizes the research efforts and provides conclusions and recommendations for

future research.

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 building

façade 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 warm 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 crystal 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 upward, 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 duration, 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

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 air 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 airborne (Houze, 1994), and are smaller than their stratiform

counterparts. Convective precipitation generally produces more intense rainfall rates

over a shorter duration than stratiform precipitation (Tokay 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).

Raindrop 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 intensity (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)

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 water 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 (1948) 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 gamma 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):

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 FCMP data acquired during Hurricane Ike. , , and

were estimated for each time history; the 50th percentiles of the parameters’ estimates

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 the distribution.

29

Figure 2-1. Influence of the modified three-parameter gamma model parameters

Willis and Tattleman (1989) expanded the gamma model by researching the large

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). The

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 normalized data:

30

(2-14)

(2-15)

(2-16)

(2-17)

(2-18)

Validation of the Willis and Tattelman (1989) 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 rain (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 accumulated 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).

31

Figure 2-2. Components of the rain intensity vector

Quantification of the Rain Deposition on the Building Façade

Methods for quantifying wind-induced wetting of the building façade were

developed by Choi (1993), Straube and Burnett (2000), and Blocken and Carmeliet

(2002) . The WDR deposition on the building is defined as the ratio of wetting on the

building at a specified location on the façade 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 façade 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 façade is the Local Intensity Factor ( ). The is obtained by integrating the

s over all hydrometeor diameters (Choi, 1994):

Wind Velocity

Driving Rain Intensity(RWDR)

Ho

riz.

Rai

nfa

ll In

ten

sity

(RH)

32

(2-20)

The wetting of the building façade 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 Affecting Rain Deposition Rate on the Building Façade

Rainfall intensity

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 the

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 more

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 Eastern United States is 274 mm/hr

for a 100 year return 5 min rain event.

33

Influence of the wind on raindrop trajectory

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 can 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 traveling through the boundary layer in the free-stream are assumed to

have a horizontal velocity component that asymptotically 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

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 valid.

As raindrops approach the building façade, 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 façade, particularly in the top quarters.

A)

0 1 2 3 4 5 6 7 8 9 10-5

-4

-3

-2

-1

0Droplet Trajectory at 70 m/s Wind Speed

Distance (m)

Dis

tance (

m)

5.0 mm

4.0 mm

3.0 mm

2.0 mm

1.0 mm

0.5 mm

0.01 mm

35

B)

C)

Figure 2-4. Trajectory of drops released at 0 m/s

A)

0 1 2 3 4 5 6 7 8 9 10-5

-4

-3

-2

-1


Distance (m)

Dis

tance (

m)

5.0 mm

4.0 mm

3.0 mm

2.0 mm

1.0 mm

0.5 mm

0.01 mm

0 10 20 30 40 50 60 70 80 90-20

-15

-10

-5


Distance (m)

Dis

tance (

m)

5.0 mm

4.0 mm

3.0 mm

2.0 mm

1.0 mm

0.5 mm

0.01 mm

0 10 20 30 40 50 60 70 80 900

10

20

30

40

50Droplet Trajectory Angle at 70 m/s Wind Speed

Distance (m)

Angle

(deg)

5.0 mm

4.0 mm

3.0 mm

2.0 mm

1.0 mm

0.5 mm

0.01 mm

95% of terminal angle

36

B)

C)

Figure 2-5. Distance required for the trajectory angle to reach 95% of the terminal angle

Terrain characteristics

Terrain characteristics affect the wetting of the building façade 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 façade; these characteristics range from ground

surface roughness dictating the global terrain exposures and overall flow conditions

(e.g., open, suburban, urban) to larger obstructions introducing local disturbances to the

0 10 20 30 40 50 60 70 80 900

10

20

30

40


Distance (m)

Angle

(deg)

5.0 mm

4.0 mm

3.0 mm

2.0 mm

1.0 mm

0.5 mm

0.01 mm


0 10 20 30 40 50 60 70 80 900

10

20

30

40


Distance (m)

Angle

(deg)

5.0 mm

4.0 mm

3.0 mm

2.0 mm

1.0 mm

0.5 mm

0.01 mm


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 mixing due to blockage and shielding effects of building obstructions at close

vicinity—causes the distribution of wetting on the building façade 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 defined 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 influences 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 the temporal and spatial deposition of rain on the building façade (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

38

upstream conditions, building orientation in the flow field, and building geometric 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 façade of large drops are less affected than those corresponding to

smaller drops (Figure 2-6, after Blocken and Carmeliet, 2002).

Figure 2-6. Deposition of smaller drops left and larger drops right (arrows indicate

decreasing drop diameter)

The effect of varying building geometry, in particular width 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 investigation 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 Façade

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-

39

empirical 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 inverse 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 coefficient. 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 the most common

models are the Straube and Burnett (2000) model and the British Standard BS EN ISO

15927-3 (2009). Both models implement:

(2-25)

where is the WDR coefficient 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 than the Straube and Burnett model, neither accurately

model the blockage effect of the tested buildings.

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 façade, is the roughness

coefficient representative of roughness of the terrain upwind of a wall (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. buildings, fences, trees, etc.)

close to and upwind of building façade (conservatively assumed to be one), and is

the wall factor which is calculated as the ratio of water reaching the building façade to

the quantity passing through an equivalent unobstructed 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. Choi (1993, 1994) proposed a method for calculating WDR deposition on

the building façade that includes: (1) calculating wind flow under steady state conditions

by solving the Navier-Stokes equations with the k-ε turbulence model, and (2)

calculating drop trajectories at every point for each raindrop size by iteratively solving

their equations of motion 2-27 through 2-29, and 2-19 through 2-20 calculating and

values at different locations of the building.

(2-27)

41

(2-28)

(2-29)

In the equations of motion (2-27 – 2-29) 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; Bitsumalak, 2009; Lopez et al., 2011).

Wind tunnel experiments have been used to characterize wetting patterns on the

façade of multiple scale models. Due their complexity and high cost, wind tunnel tests of

WDR have been limited. Major difficulties encountered during 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).

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 rate 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 Engineering

Limited (2002), Florida International University (Bitsuamlak et al., 2009), and the

University of Florida (Masters et al.,2008; Salzano et 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 protocols, 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 was 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 performance evaluation

of the water penetration resistance of building products and systems.

43

Measurement of Wind-Driven Rain

Instrumentation

In the earliest studies, WDR measurements were made using directional

pluviometers (known as vectopluviometers) and wall mounted 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, respectively—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 WDR data that has been used for the development of WDR maps

(Lacy,1965), as a tool for estimating WDR deposition on building façades, 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 façade. 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 facilitated 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 measurement 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 can vary from the ground level RSD, and the RSD

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 been investigated by Marshall et al. (1947), Wilson and Brandes (1979), and

Medlin et al. (2007).

Disdrometers can accurately measure ground level 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 disdrometers, 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 interpreted

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 for 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 obstructed 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 disdrometers are also more accurate than impact disdrometers

in measuring particles less than 0.7mm (Loffler-Mang and Joss, 2000).

45

Limitations of optical disdrometers

Currently, optical disdrometers are the instrument of choice for 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.

Fringe effects refer to the partial measurement of a drop’s diameter as the drop

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 error 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 travelling 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 terminal 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 .

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 becomes 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, depending 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).

Many attempts have been made to mitigate these effects (Donnadieu, 1980;

Hauser et al., 1984; Grossklaus et al., 1998; Nespor et al., 2000; Tokay et al., 2001).

Disdrometers have been developed 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 measure 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 from 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 employing quality

control algorithms that filter data according to the measured drop velocities and whether

47

or not they fall within a velocity threshold based on the work done by Gunn and Kinzer

(1949). These quality control 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 certain

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) disdrometer (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 instrument 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

measurements, reporting precipitation at ten and one second intervals, respectively.

The instruments are described in the following section.

Instrumentation Used In This Study

OTT PARSIVEL optical disdrometer

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, diameter, 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

48

voltage from the steady state 5 V. The voltage time history is inverted, amplified,

filtered, and the DC component is removed (Loffler-Mang and Joss, 2000) such that the

diameter of the particle is estimated from voltage drop (ΔV). The duration of light

absence (Δt), from when the particle enters and exits the laser 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 digitally 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 1.000

22 6.500 1.000

23 7.500 1.000

24 8.500 1.000

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 classes

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

50

Figure 2-7. PARSIVEL drop diameter and velocity determination

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 measurement and output

0.30 mm drops (class 3)

traveling 0.34 m/s (class 3)

0.70 mm drops (class 6)

traveling 0.6 m/s (class 7)

51

Droplet Measurement Technologies Precipitation Imaging Probe

The Droplet Measurement Technologies (DMT) Precipitation Imaging Probe (PIP)

is a 2-D optical disdrometer 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 area 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)

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, simulation

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.

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. To the author’s knowledge, this is the first known effort to successfully reduce

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 organized 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

travelling 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

instrument 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 Chapter 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

54

sensor was greater than that of the sensor in the stationary, horizontally aligned,

configuration (the proof is shown 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.

Raindrop Size Distribution 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 was 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 using 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 dish 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 SLR 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

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 Parmer #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 each 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 medium 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

1.00 mm

10x Magnification

Reference1.0 cm x 1.0

cmArea

0.15 mm

30x Magnification

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 measurements 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 results 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 recorded indicating that incorrect drop diameter measurements may

0 0.5 1 1.5 2 2.5 3 3.5 410

-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n (

mm

-1 m

-3)

PARSIVEL RSDs

Mean PARSIVEL RSD

RSDs from Algorithm

Mean RSD from Algorithm

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%, yielding incorrect RSD

measurements. The effect of oblique trajectory angles was further investigated in winds

generated by the UF Windstorm Simulator.

Table 3-1. Trajectory angles of multiple diameter drops in multiple wind speeds

Drop Diameter

(mm) Vt(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°

θ

Bearing

Direction

6.1m

Drop

Height

Laser

Band

58

Figure 3-4. PARSIVEL measured diameters at multiple trajectory angles

0 1 2 3 405

101 mm Bearing measured at 90 deg

Measured Diameter (mm)

Co

un

t

0 1 2 3 405



Co

un

t

0 1 2 3 405



Co

un

t

0 1 2 3 40

1020



Co

un

t

0 1 2 3 405



Co

un

t

0 1 2 3 405



Co

un

t

0 1 2 3 405



Co

un

t

0 1 2 3 405



Co

un

t

0 1 2 3 405



Co

un

t

0 1 2 3 405



Co

un

t0 1 2 3 4

05



Co

un

t

0 1 2 3 405



Co

un

t

0 1 2 3 40



Co

un

t

0 1 2 3 405



Co

un

t

0 1 2 3 40

1020



Co

un

t

0 1 2 3 405



Co

un

t

0 1 2 3 405



Co

un

t

0 1 2 3 405



Co

un

t

0 1 2 3 405



Co

un

t

0 1 2 3 405



Co

un

t

0 1 2 3 405



Co

un

t

59

Figure 3-5. PARSIVEL measured velocities at multiple trajectory angles (dashed line indicates theoretical velocity)

0 5 100

1020


Measured Velocity (m/s)

Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020



Co

un

t

0 5 100

1020


Measured Velocity (m/s)C

ou

nt

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 nozzle 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, at 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

which 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) proposed, 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.

61

Table 3-2. Steady wind test matrix

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 (photo courtesy of author)

62

Figure 3-7. PARSIVEL measured raindrop size distributions for the tests in steady wind flow

0 2 4 610

-2

100

102

104

106

Diameter (mm)

U =

8.9

m/s

Co

nce

ntr

atio

n (

mm

-1m

-3) R

wdr ~ 400 mm/hr

0 2 4 610

-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n (

mm

-1m

-3)

Measured RSDR

wdr ~ 475 mm/hr

0 2 4 610

-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n (

mm

-1m

-3) R

wdr ~ 550 mm/hr

0 2 4 610

-2

100

102

104

106

Diameter (mm)

U =

17

.9 m

/s

Co

nce

ntr

atio

n (

mm

-1m

-3)

0 2 4 610

-2

100

102

104

106

Diameter (mm)

U =

35

.8 m

/s

Co

nce

ntr

atio

n (

mm

-1m

-3)

0 2 4 610

-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n (

mm

-1m

-3)

0 2 4 610

-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n (

mm

-1m

-3)

0 2 4 610

-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n (

mm

-1m

-3)

0 2 4 610

-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n (

mm

-1m

-3)

0 deg 22.5 deg 45 deg PIP

63

Figure 3-8. PARSIVEL measured drop diameters and velocities for the tests in steady wind flow

Diameter (mm)

8.9

m/s

Ve

locity (

m/s

)0 deg

0 1 2 30

5

10

15

0

100

200

300

400

500

Diameter (mm)

Ve

locity (

m/s

)

BETE WL1-1/2 Nozzle Grid at ~ 400 mm/hr22.5 deg

0 1 2 30

5

10

15

Diameter (mm)

Ve

locity (

m/s

)

45 deg

0 1 2 30

5

10

15

Diameter (mm)

17

.9 m

/sV

elo

city (

m/s

)

0 1 2 30

5

10

15

Diameter (mm)

35

.8 m

/sV

elo

city (

m/s

)

0 1 2 30

5

10

15

0

100

200

300

400

500

0

100

200

300

400

500

0

100

200

300

400

500

Diameter (mm)

Ve

locity (

m/s

)

0 1 2 30

5

10

15

0

100

200

300

400

500

Diameter (mm)

Ve

locity (

m/s

)

0 1 2 30

5

10

15

0

100

200

300

400

500

0

100

200

300

400

500

Diameter (mm)

Ve

locity (

m/s

)

0 1 2 30

5

10

15

0

100

200

300

400

500

Diameter (mm)

Ve

locity (

m/s

)

0 1 2 30

5

10

15

0

100

200

300

400

500

64

Articulating Instrumentation Platforms

An instrumentation 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 platforms 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 control 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 PARSIVEL 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 Labview 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

65

Figure 3-9. Articulating instrument platform

RM Young sonic anemometer

Wind velocity and 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 anemometers 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 transducers. 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.

66

Articulating support structure

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 interpreted 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. Articulating 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

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 was found

to be while the moment required to overturn the instrument was found to be

.

Figure 3-11. Drag coefficient of a cylinder dependant of Reynolds number of flow (Adapted from R Panton, Incompressible Flow 3rd ed.)

Data acquisition system

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 operating on the National Instruments LabView 8.5

platform. The control and data diagram is shown in Figure 3-12. The system is also

68

capable of streaming summary data via a cellular connection, should future needs

warrant an upgrade.

Figure 3-12. Instrument platform control and data diagram

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

Digita

l Signa

lE

levatio

n A

ngle

Control Algorithm

Win

d S

pee

dA

nal

og

0-1

0V

Win

d D

irec

tio

nA

nal

og

0-1

0V

Data Logger/ National Instruments Labview 8.5

Software

Digital SignalWind Data

Digital Sign

al

Azim

uth

Digital Sign

al

Rain

Da

ta

Elevation Angle

Stepper Motor

Azimuth

Stepper Motor

69

data and power cables are connected to ports on the enclosure box and the software is

set to operate.

The use of a collapsible tripod that assembles in the field permitted the system’s

portability and rapid deployment. During the VORTEX2 deployment the instrument was

assembled, operational, and 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 pose 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 outputs 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 instrumentation 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 vector. 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, respectively. The

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 drops at low speed) are mostly excluded from the analyzed

dataset.

A)

0 5 10 15 200

2

4

6

8

10

12

14

16

18

20

Diameter (mm/hr)

Ve

locity (

m/s

)

U = 0 m/s

Considered D-V Combinations

VT

VR

Observed

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 anemometers (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 redundant 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

height. 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

0 5 10 15 200

2

4

6

8

10

12

14

16

18

20

Diameter (mm/hr)

Velo

city (

m/s

)

U = 10 m/s

Considered D-V Combinations

VT

VR

Observed

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 for 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 Precipitation Imaging Probe turret system and Gill anemometers at 5.0

m (photo courtesy of author)

73

Comparison of Collocated Stationary and Articulating Instruments in a Supercell Thunderstorm

Data from a supercell thunderstorm 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 measured 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 lead to incorrect estimates of all

rainfall parameters in the stationary instrumentation platforms (rainfall intensity - Figure

3-16 and reflectivity - Figure 3-17). The articulating instrument platform did not exhibit

similar jumps in the data, indicating reliable measurements.

A) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

0

0.2

0.4

0.6

0.8

1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

0.2

0.4

0.6

0.8

1

Time

Dro

p D

iam

ete

r (m

m)

Measured RSD on Jun 10, 2010

01:17Z 01:25Z 01:33Z 01:41Z 01:49Z 01:58Z

1

2

3

4

5

6

7

8

10-1

100

101

102

103

74

B)

C)

Figure 3-15. Measured raindrop size distribution by stationary instrumentations (A) and articulating instrumentation (B)

Figure 3-16. Comparison of rainfall intensities measured by stationary and articulating instrument platforms

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

0.2

0.4

0.6

0.8

1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

0.2

0.4

0.6

0.8

1

Time

Dro

p D

iam

ete

r (m

m)

Measured RSD on Jun 10, 2010

01:17Z 01:25Z 01:33Z 01:41Z 01:49Z 01:58Z

1

2

3

4

5

6

7

8

10-1

100

101

102

103

01:17Z 01:25Z 01:33Z 01:41Z 01:49Z 01:58Z0

5

10

15

Time

Win

d S

pe

ed

(m/s

)

01:17Z 01:25Z 01:33Z 01:41Z 01:49Z 01:58Z0

200

400

600

800

1000

1200

Time

Ra

infa

ll r

ate

(mm

/hr)

Measured Rainfall Rate Comparison of Data Collected on Jun 10, 2010

Articulating Instrument

Stationary Instrument

75

Figure 3-17. Comparison of estimated reflectivity by stationary and articulating

instrument platforms

Summary

This chapter presented the known inaccuracies in RSD characterization caused by

strong winds. As a result, an articulating instrument platform was designed, prototyped,

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 results of data

collected during the VORTEX2 and FCMP field campaigns.

01:17Z 01:25Z 01:33Z 01:41Z 01:49Z 01:58Z0

20

40

60

80

100

Time

Re

fle

ctivity(d

BZ

)

Reflectivity Comparison of Data Collected on Jun 10, 2010

Articulating Instrument

Stationary Instrument

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 (VORTEX2) 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 six 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 multiple wind velocities, (5)

calculation of the peak to mean ratios of rainfall intensity ( ), and (6) comparison of

measured rain data to WSR-88D 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 to investigate tornado genesis,

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 instruments 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 scientists in better

understanding tornadic behavior and improving tornado forecasting.

Deployment details

As part of VORTEX2, a disdrometer team was tasked with the repeated

deployment of eight instrument stations (articulating and stationary) for the collection 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 activity. 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 consists 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

78

3). The data 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 Tornadoes Experiment 2 deployment

details

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 UF6 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

25-May 0005-0100 Ogallala, NE CU1 UF5 UF6 UF7

*29-May 0015-0140 Thedford, NE CU1 UF1*UF3* UF4 UF5 UF6 UF7

*02-Jun 2255-2355 Benkelman, NE CU1 UF1*UF3* UF5 UF6

*03-Jun 0020-0100 Wagner, SD UF1*UF3* UF5 UF6

*05-Jun 2330-0010 Des Moines, IA CU1 UF1*UF3* UF5 UF6

*06-Jun 2220-0210 Ogallala, NE CU1 UF1* UF5 UF6

*06-Jun 0110-0210 Ogallala, NE CU1 UF1* UF5 UF6

*07-Jun 2335-0030 Scottsbluff, NE CU1 UF1*UF3* UF5 UF6 UF7

08-Jun 0100-0200 Scottsbluff, NE CU1 UF5 UF6 UF7

*09-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 UF5 UF6 UF7

*13-Jun 2250-2340 Darrouzett, TX CU1 UF5 UF6 UF7

*14-Jun 2000-2042 Post, TX CU1 UF5 UF6 UF7

79

Figure 4-1. VORTEX2 instrument deployment

Figure 4-2. VORTEX2 data 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

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 data 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 details

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 the 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

81

exposure for the weather station was primarily open to the east and suburban

elsewhere. 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 Rico. 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 northward 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 continued

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

tropical 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.

To compare data collected by weather station T-3 to data gathered using the

articulating instrument platform, 10 second segmental averages of disdrometer data and

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 longitudinal 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 ( ),

mean 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 of 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 related 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 analysis.

Calculated PDFs of VORTEX2 data indicate that and (Figure 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

83

indicates 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 negative 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 PDFs 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 speed. 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 higher 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 turbulence 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.

84

Figure 4-3. Effect of longitudinal wind velocity on drop diameter observed in VORTEX2 data

85

Figure 4-4. Effect of longitudinal turbulence intensity on drop diameter observed in VORTEX2 data

86

Figure 4-5. Effect of lateral turbulence intensity on drop diameter observed in VORTEX2 data

87

Figure 4-6. Effect of longitudinal wind velocity on drop diameter observed in FCMP data

88

Figure 4-7. Effect of longitudinal turbulence intensity on drop diameter observed in FCMP data

89

Figure 4-8. Effect of lateral turbulence intensity on drop diameter observed in FCMP data

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

into 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

parameters 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 of , , 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 made 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-1observed from typhoon data collected by Chang et al.

(2009); 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 necessary to make an assessment.

91

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

0 10 20 3010

0

105

1010

U (m/s)

N0 (

mm

-1 m

-3)

0 0.2 0.410

0

105

1010

TIU

N0 (

mm

-1 m

-3)

0 0.2 0.410

0

105

1010

TIV

N0 (

mm

-1 m

-3)

0 20 40 6010

0

105

1010

R (mm/hr)

N0 (

mm

-1 m

-3)

0 10 20 30-10

0

10

20

U (m/s)

0 0.2 0.4-10

0

10

20

TIU

0 0.2 0.4-10

0

10

20

TIV

0 20 40 60-10

0

10

20

R (mm/hr)

0 10 20 30-10

0

10

20

U (m/s)

(

mm

-1)

0 0.2 0.4-10

0

10

20

TIU

(

mm

-1)

0 0.2 0.4-10

0

10

20

TIV

(

mm

-1)

0 20 40 60-10

0

10

20

R (mm/hr)

(

mm

-1)

Data in U <= 15 m/s

92

Figure 4-10. FCMP Hurricane Ike and Irene gamma parameters observed in multiple wind conditions and rainfall

intensities

0 10 20 3010

0

105

1010

U (m/s)

N0 (

mm

-1 m

-3)

0 0.2 0.410

0

105

1010

TIU

N0 (

mm

-1 m

-3)

0 0.2 0.410

0

105

1010

TIV

N0 (

mm

-1 m

-3)

0 20 40 6010

0

105

1010

R (mm/hr)

N0 (

mm

-1 m

-3)

0 10 20 30-10

0

10

20

U (m/s)

0 0.2 0.4-10

0

10

20

TIU

0 0.2 0.4-10

0

10

20

TIV

0 20 40 60-10

0

10

20

R (mm/hr)

0 10 20 30-10

0

10

20

U (m/s)

(

mm

-1)

0 0.2 0.4-10

0

10

20

TIU

(

mm

-1)

0 0.2 0.4-10

0

10

20

TIV

(

mm

-1)

0 20 40 60-10

0

10

20

R (mm/hr)

(

mm

-1)

Data in U <= 15 m/s

Data in U > 15 m/s

93

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

0 10 20 3010

0

105

1010

U (m/s)

N0 (

mm

-1 m

-3)

0 0.2 0.410

0

105

1010

TIU

N0 (

mm

-1 m

-3)

0 0.2 0.410

0

105

1010

TIV

N0 (

mm

-1 m

-3)

0 20 40 6010

0

105

1010

R (mm/hr)

N0 (

mm

-1 m

-3)

0 10 20 30-10

0

10

20

U (m/s)

0 0.2 0.4-10

0

10

20

TIU

0 0.2 0.4-10

0

10

20

TIV

0 20 40 60-10

0

10

20

R (mm/hr)

0 10 20 30-10

0

10

20

U (m/s)

(

mm

-1)

0 0.2 0.4-10

0

10

20

TIU

(

mm

-1)

0 0.2 0.4-10

0

10

20

TIV

(

mm

-1)

0 20 40 60-10

0

10

20

R (mm/hr)

(

mm

-1)

Data in U <= 15 m/s

Data in U > 15 m/s

94

Figure 4-12. Observed shape slope relation

0 2 4 6 8 10 12 14 16 18-4

-2

0

2

4

6

8

10

12

14

(mm-1

)

Data Collected in U <= 15 m/s

Data Collected in U > 15 m/s

Zhang et al. 2003

95

Comparison of Raindrop Size Distribution Models to Measured Raindrop Size Distribution Data in Multiple Wind Velocities

RSD models serve as a simple method for acquiring 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 square error ( ) values were calculated

and listed in Figure 4-16 -Figure 4-18. Each value was computed as follows:

(4-4)

where is the measured RSD and is the fit RSD. The values confirm that

there was no significant difference 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.

96

Figure 4-13. Model raindrop size distribution and measured raindrop size distribution

comparisons (N indicates the number of averaged records)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 2597

NFCMP

= 64378

U = 0m/s - 15 m/s

R =

0 -

20 m

m/h

r

ND (

mm

-1 m

3)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 4

NFCMP

= 15822

U = 15m/s +

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 631

NFCMP

= 1039

R =

20 -

40 m

m/h

r

ND (

mm

-1 m

3)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 2

NFCMP

= 2640

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 336

NFCMP

= 335

R =

40 -

60 m

m/h

r

ND (

mm

-1 m

3)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 1

NFCMP

= 1088

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 232

NFCMP

= 172

R =

60 -

80 m

m/h

r

ND (

mm

-1 m

3)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 3

NFCMP

= 570

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 464

NFCMP

= 134

R =

80 m

m/h

r +

ND (

mm

-1 m

3)

Diameter (mm)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 16

NFCMP

= 585

Diameter (mm)

VORTEX2 FCMP Best Gamma Willis and Tattleman Marshall Palmer

97

Figure 4-14. Model raindrop size distribution and measured raindrop size distribution

comparisons (N indicates the number of averaged records)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 1177

NFCMP

= 52252

TIU = 0 - 0.2

R =

0 -

20 m

m/h

r

ND (

mm

-1 m

3)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 1424

NFCMP

= 27948

TIU = 0.2 +

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 247

NFCMP

= 2754

R =

20 -

40 m

m/h

r

ND (

mm

-1 m

3)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 386

NFCMP

= 925

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 151

NFCMP

= 1061

R =

40 -

60 m

m/h

r

ND (

mm

-1 m

3)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 186

NFCMP

= 362

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 113

NFCMP

= 551

R =

60 -

80 m

m/h

r

ND (

mm

-1 m

3)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 122

NFCMP

= 191

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 265

NFCMP

= 564

R =

80 m

m/h

r +

ND (

mm

-1 m

3)

Diameter (mm)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 215

NFCMP

= 155

Diameter (mm)


98

Figure 4-15. Model raindrop size distribution and measured raindrop size distribution comparisons (N indicates the number of averaged records)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 1520

NFCMP

= 78337

TIV

= 0 - 0.2

R =

0 -

20 m

m/h

r

ND (

mm

-1 m

3)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 1081

NFCMP

= 1863

TIV

= 0.2 +

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 365

NFCMP

= 3659

R =

20 -

40 m

m/h

r

ND (

mm

-1 m

3)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 268

NFCMP

= 20

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 211

NFCMP

= 1413

R =

40 -

60 m

m/h

r

ND (

mm

-1 m

3)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 126

NFCMP

= 10

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 158

NFCMP

= 742

R =

60 -

80 m

m/h

r

ND (

mm

-1 m

3)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 77

NFCMP

= 0

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 353

NFCMP

= 717

R =

80 m

m/h

r +

ND (

mm

-1 m

3)

Diameter (mm)

0 1 2 3 4 510

-2

100

102

104

106

NVORTEX2

= 127

NFCMP

= 2

Diameter (mm)


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 distribution models stratified by U and R


Model



TIU TIU TIU TIU TIU TIU TIU TIU

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 distribution models stratified by TIU and R

100


Model



TIV TIV TIV TIV TIV TIV TIV TIV

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 2.05 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 0.62 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 TIV and R

101

Peak to Mean Ratio of Rainfall Intensities

Currently, the widely accepted standard used to estimate the design wetting rate

on building façades (BSI EN ISO 15927-3) is based on a minimum of ten years of hourly

and data. The design free-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 stringent 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 figure 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

102

t-test was conducted to compare the datasets at one and ten seconds (VORTEX2 and

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 WDRFS

R 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

103

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

10 20 30 40 50 600

1

2

3N = 61

Stratiform (< 40 dBZ)

U <

15 m

/s

10 20 30 40 50 600

1

2

3N = 22

Convective (> 40 dBZ)

10 20 30 40 50 600

1

2

3N = 90

Stratiform and Convective

10 20 30 40 50 600

1

2

3N = 0

U >

15 m

/s

10 20 30 40 50 600

1

2

3N = 0

10 20 30 40 50 600

1

2

3N = 0

10 20 30 40 50 600

1

2

3N = 61

All

U

Duration (sec)

10 20 30 40 50 600

1

2

3N = 22

Duration (sec)

10 20 30 40 50 600

1

2

3N = 90

Duration (sec)

25th Percentile 50th Percentile 75th Percentile

10 20 30 40 50 600

1

2

3N = 61


U <

15 m

/s

10 20 30 40 50 600

1

2

3N = 22


10 20 30 40 50 600

1

2

3N = 90


10 20 30 40 50 600

1

2

3N = 0

U >

15 m

/s10 20 30 40 50 60

0

1

2

3N = 0

10 20 30 40 50 600

1

2

3N = 0

10 20 30 40 50 600

1

2

3N = 61

All

U

Duration (sec)

10 20 30 40 50 600

1

2

3N = 22

Duration (sec)

10 20 30 40 50 600

1

2

3N = 90

Duration (sec)


104

Figure 4-20. Peak to mean ratios of FCMP data (N indicates number of one minute segments)

0 20 40 600

2

4

6

8

10N = 470


U <

15 m

/s

0 20 40 600

2

4

6

8

10N = 11


0 20 40 600

2

4

6

8

10N = 629


0 20 40 600

2

4

6

8

10N = 227

U >

15 m

/s

0 20 40 600

2

4

6

8

10N = 101

0 20 40 600

2

4

6

8

10N = 394

0 20 40 600

2

4

6

8

10N = 706

All

U

Duration (sec)

0 20 40 600

2

4

6

8

10N = 112

Duration (sec)

0 20 40 600

2

4

6

8

10N = 1036

Duration (sec)


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 consisted 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 intensity 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 compared 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 good 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

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 and 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 relationship 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

3

6

m

mm Z (dBZ) R2

Best Fit Curve

Rosenfeld tropical

Default WSR-88D

Marshall/Palmer

Figure 4-21. Comparison of disdrometer estimated and radar measured reflectivity

03:00Z 06:00Z 09:00Z 12:00Z 15:00Z 18:00Z0

10

20

30

40

50

Time (UTC)

Z (

dB

Z)

KGHX Data

FCMP Data

107

Figure 4-22. Comparison of disdrometer measured and radar estimated rainfall intensity

Figure 4-23. Observed Z-R relationship

03:00Z 06:00Z 09:00Z 12:00Z 15:00Z 18:00Z0

10

20

30

40

50

Time (UTC)

R (

mm

/hr)

KGHX Data

FCMP Data

0 10 20 30 40 500

10

20

30

40

50

60

Reflectivity (

dB

Z)

R (mm/hr)

FCMP Data

Best fitZ = 15.5 + 17.4log(R)

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 the 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. Observations 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.

0 10 20 30 40 50 600

10

20

30

40

50

60

70

80

R (

mm

/hr)

Reflectivity (dBZ)

Measured Data

Best Fit Curve

Default Z-R

Rosenfeld

Marshall and Palmer

109

CHAPTER 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

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 the 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 rate 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 tropical 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 determine 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

110

and in steady wind to determine if the presence of wind caused the changes in the

RSD. The differences were found 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 about 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 mm/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.

As discussed in Chapter 2, rain ‘intensity’ 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

111

divided by the horizontal area 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

deposition rate ( ), which is the specified wetting rate on the building façade (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.

The 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.

2-24) 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 selected 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 assuming the wind speed U = 60 m/s. Italicized values are estimates.

Source RH RWDR RF = 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

112

single nozzle (L3T-1) by the grid cell area (L2), 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 and 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-shelf 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 protocol 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 performed to quantify the

variability of the spray using the apparatus shown in Figure 5-1a. A single nozzle was

aligned perpendicularly to the center of a vertically oriented collection chamber with 12

receivers. Each 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

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 5-1B. 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 experimental 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

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. Three 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 sample 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 in 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 is necessary to get reliable

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 was time

and cost prohibitive given the fuel, water, and labor resources necessary to operate the

wind generator.

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 distributions measured in stagnant air and in a

steady wind by the PARSIVEL (left) and PIP (right, BETE WL 1 1/2 at 138 kPa)

Characterization of the Raindrop Size Distribution of a Spay Nozzles in Stagnant Air: A Proxy for Full-Scale Testing

Specimen Matrix

The 12 full cone hydraulic nozzles 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 nozzles evaluated in this study

BETE Lechler Steinen Spray Systems

WL 1 ½ 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

0 1 2 3 410

-2

100

102

104

106

Diameter (mm)

Concentr

ation (

mm

-1 m

-3)

U = 0 m/s R

WDR = 509 mm/hr

U = 8.9 m/s, RWDR

= 172 mm/hr

U = 17.9 m/s, RWDR

= 155 mm/hr

U = 35.8 m/s, RWDR

= 36 mm/hr

0 1 2 3 410

-2

100

102

104

106

Diameter (mm)

Concentr

ation (

mm

-1 m

-3)

U = 0 m/s R

WDR = 509 mm/hr

U = 8.9 m/s, RWDR

= 977 mm/hr

U = 17.9 m/s, RWDR

= 1020 mm/hr

U = 35.8 m/s, RWDR

= 1050 mm/hr

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 test 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

x x x x x x x xxxxxxxx

xxxxxxxx x x x x x x

3.05 m

Tests @ 0.31 m

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 angle 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<Re< 800, Schiller and Nauman (1933) define the drag coefficient and force as:

(5-2)

(5-3)

For 800<Re<105, Clift and Gauvin (1970) define the drag coefficient and force as:

(5-4)

(5-5)

where is the drag coefficient, is the force vector, is the Reynolds number,

is air density, is water density, is dynamic viscosity of air, is the drop

mass, is gravitational acceleration, is the cross-sectional area of each particle,

is the wind velocity vector, and is the velocity vector of the particles. The

acceleration ( ), velocity( ), and position ( ) are iteratively solved for each time

step as follows:

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 designs)

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 for

all nozzles was similar.

119

Figure 5-4. Count of drops radially outward from nozzle centerline

0 1 2 3 40

1

2

3x 10

4

r (m)

# D

rop

s

D = 0.31 mm

0 1 2 3 40

2

4

6x 10

4

r (m)

# D

rop

s

D = 0.44 mm

0 1 2 3 40

2

4

6x 10

4

r (m)

# D

rop

s

D = 0.56 mm

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.69 mm

0 1 2 3 40

1

2x 10

4

r (m)

# D

rop

s

D = 0.81 mm

0 1 2 3 40

5000

10000

r (m)

# D

rop

s

D = 0.94 mm

0 1 2 3 40

5000

r (m)

# D

rop

s

D = 1.1 mm

0 1 2 3 40

2000

4000

r (m)

# D

rop

s

D = 1.2 mm

0 1 2 3 40

2000

4000

r (m)

# D

rop

s

D = 1.4 mm

0 1 2 3 40

1000

2000

r (m)

# D

rop

s

D = 1.6 mm

0 1 2 3 40

500

1000

1500

r (m)

# D

rop

s

D = 1.9 mm

0 1 2 3 40

500

1000

r (m)

# D

rop

s

D = 2.1 mm

0 1 2 3 40

200

400

r (m)

# D

rop

s

D = 2.4 mm

0 1 2 3 40

200

400

600

r (m)

# D

rop

s

D = 2.8 mm

0 1 2 3 40

50

100

150

r (m)

# D

rop

sD = 3.3 mm

0 1 2 3 40

10

20

30

r (m)

# D

rop

s

D = 3.8 mm

120

Figure 5-5. Determining initial velocity using high speed footage (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:

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. Raindrop size distribution of BETE WL3 in stagnant air conditions

0 1 2 3 4 510

-4

10-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n, N

D (

mm

-1 m

-3)

BETE WL3

Other Nozzles

Best (1950)

122

Validation of the Wind-Driven Rain 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 façade will be

approximately 203 mm/hr and the RSD will be representative of natural conditions.

123

Figure 5-7. Instrument arrangement at the Insurance Institute for Business & Home Safety Research Center (photo courtesy of author)

124

Figure 5-8. Measured raindrop size distributions at multiple heights and wind velocities

0 1 2 3 4 5

101

104

107

U = 8.9 m/sZ

= 1

.8 m

0 1 2 3 4 5

101

104

107

U = 17.9 m/s

0 1 2 3 4 5

101

104

107

U = 26.8 m/s

0 1 2 3 4 5

101

104

107

Z =

2.7

m

0 1 2 3 4 5

101

104

107

0 1 2 3 4 5

101

104

107

0 1 2 3 4 5

101

104

107

Z =

3.7

m

0 1 2 3 4 5

101

104

107

0 1 2 3 4 5

101

104

107

0 1 2 3 4 5

101

104

107

Z =

5.2

m

0 1 2 3 4 5

101

104

107

0 1 2 3 4 5

101

104

107

0 1 2 3 4 5

101

104

107

Z =

6.0

m

Diameter (mm)0 1 2 3 4 5

101

104

107

Diameter (mm)0 1 2 3 4 5

101

104

107

Diameter (mm)

Parsivel Measurement PIP MeasurementBest (1950) ModelBased on 509 mm/hr

125

Figure 5-9. Comparison of measured raindrop size distributions and Best (1950) model

Summary

This chapter presented the design of a realistic wind-driven 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 Research 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.

0 1 2 3 4 510

0

101

102

103

104

105

106

Co

nce

ntr

atio

n (

mm

-1 m

-3)

Diameter (mm)

Parsivel Mean

PIP Mean

Best (1950) Model for R = 509 mm/hr

126

CHAPTER 6

SUMMARY, CONCLUSIONS,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 simulation system.

Characterization of Extreme Wind-Driven Rain 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 measurement 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 was 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 size 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

127

processing algorithm that was created to count the number of droplets captured in an oil

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 droplets 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 velocities 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 exposed 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 velocities.

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

128

instrumentation becomes compromised; while the data from the articulating

instrumentation exhibited no wind induced error.

Conclusions from Field Data Results

The simultaneous wind and rain data gathered by the articulating instruments, to

the author’s knowledge, has been the first attempt at characterizing ground level free-

stream RSDs 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

velocities, 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 diameter

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 intensities 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.

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, turbulence 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 longitudinal

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 observed 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 Raindrop Size Distribution 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

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 relationship, 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 Rosenfeld 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 increase 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 than using hourly data. For

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 intensities 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 Wind-Driven Rain System

The Insurance Institute for Business & Home Safety tasked 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

methodology consisted of characterizing the nozzles’ 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

whose manufacturers’ specifications indicate 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 characterize 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 given 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

132

taken at the facility with the PARSIVEL 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 for 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 observations. 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 agencies 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, utilizing 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 that 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 instrumentation platforms to other applications and

decrease the occurrence of, otherwise unknown, incorrectly measured data.

133

PARSIVEL validation results from this research project indicate that the error

introduced by oblique trajectory angles manifest as incorrect 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 measurement 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

simulates 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 component 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.

Current numerical WDR models make a Stoke’s flow assumption (Re << 1) to

determine the driving forces of particles moving through space. Implementing a wind

field determined by a CFD software package to the developed trajectory model will

result in a new Reynold’s number dependent WDR model. Such a model may yield

better results and warrants further research.

134

Recommendations for the Morphological Image Processing 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;

however, it may be a useful tool for applications where the drop size distribution is

known to remain constant or for instrumentation validation.

If the morphologic image processing algorithm and Reynold’s number dependent

trajectory model are made available to the public, they may be used by nozzle

manufacturers and designers for the characterization of nozzles and for the design of

custom water delivery applications. To the author’s knowledge, no such design tool is

currently available.

135

APPENDIX

Theoretical Proof of Greater Accuracy from 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 stationary

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 area (m2) VTD = Terminal velocity of droplets of diameter D (m/s)

U = Longitudinal wind velocity (m/s)

VRD = Resultant velocity of droplets of diameter D (m/s)

CD = Number density of droplets of diameter D (m-1)

nDΦ = Number of droplets of diameter D having trajectory Φ collected per sec (sec-1)

VS = Sample volume (m3)

MDΦ = Total number of droplets having trajectory Φ collected over time T

ND = Number concentration of droplets of diameter D

Case 3 Case 2 Case 1

136

Stationary instrument (Case 1):

If we collect data for Time T and say total number of droplets = MDΦ:

We know:

Thus:

Solve for concentration of droplets:

Given:

Thus we find that the sample volume for a stationary instrument is:

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:

Check for Φ = 90 (Case where U = 0):

Checks

Say :

If we collect data for Time T and say total number of droplets = MD,Φ:

Solve for concentration of droplets:

Thus we find that the sample volume for an articulating instrument is:

138

Compare Instrumentation platforms:

Given we know:

for stationary instrumentation

for articulating instrumentation

K can vary from U to VT,D . If the tracking algorithm is accurate, in high winds K

approaches U and VT,D in no wind


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:

Ф as the tracking algorithm calculates it


for articulating instrumentation

Thus the sample volume of articulating instrument is greater than twice as large at U =

10 m/s. Similarly the ratio of the sample volumes can be calculated for multiple wind velocities:

U V T , (1.2 m m ) φ K V S ratio

0 4.64 1.57 4.64 1.00

5 4.64 0.75 6.82 1.47

10 4.64 0.43 11.02 2.38

15 4.64 0.30 15.70 3.38

20 4.64 0.23 20.53 4.42

25 4.64 0.18 25.43 5.48

30 4.64 0.15 30.36 6.54

139

Nozzle Selection

Nozzles are used in a wide range of applications, including evaporative cooling,

gas conditioning, fire 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 droplets (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

140

Figure A-2. Types of nozzles and drop size relationship

The effect of each variable is as follows:

Nozzle types, which can be hydraulic or air assisted. Air 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 drop 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.

141

Measured Diameter – Wind Relationships

Hurricane Ike Data

144

Hurricane Irene (Beaumont) Data

147

Hurricane Irene (Deal) Data

150

Florida Coastal Monitoring Program Hurricane Ike Data

153

Verification of the Origins of Rotation in Tornadoes Experiment 2 Articulating Instrument Platform Data

210

Measured Nozzle Characteristics

Uniformity

BETE WL-1

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.31 mm

0 1 2 3 40

5

10x 10

4

r (m)

# D

rop

s

D = 0.44 mm

0 1 2 3 40

5

10x 10

4

r (m)

# D

rop

s

D = 0.56 mm

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.69 mm

0 1 2 3 40

1

2x 10

4

r (m)

# D

rop

s

D = 0.81 mm

0 1 2 3 40

5000

10000

r (m)

# D

rop

s

D = 0.94 mm

0 1 2 3 40

1000

2000

r (m)

# D

rop

s

D = 1.1 mm

0 1 2 3 40

200

400

r (m)

# D

rop

s

D = 1.2 mm

0 1 2 3 40

200

400

r (m)

# D

rop

s

D = 1.4 mm

0 1 2 3 40

100

200

r (m)

# D

rop

s

D = 1.6 mm

0 1 2 3 40

20

40

r (m)

# D

rop

s

D = 1.9 mm

0 1 2 3 40

5

10

r (m)

# D

rop

s

D = 2.1 mm

0 1 2 3 40

0.5

1

r (m)

# D

rop

s

D = 2.4 mm

0 1 2 3 4-1

0

1

r (m)

# D

rop

s

D = 2.8 mm

0 1 2 3 4-1

0

1

r (m)

# D

rop

s

D = 3.3 mm

0 1 2 3 4-1

0

1

r (m)

# D

rop

s

D = 3.8 mm

211

BETE WL-2

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.31 mm

0 1 2 3 40

5

10x 10

4

r (m)

# D

rop

s

D = 0.44 mm

0 1 2 3 40

5

10x 10

4

r (m)

# D

rop

s

D = 0.56 mm

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.69 mm

0 1 2 3 40

1

2x 10

4

r (m)

# D

rop

s

D = 0.81 mm

0 1 2 3 40

5000

10000

r (m)

# D

rop

sD = 0.94 mm

0 1 2 3 40

2000

4000

r (m)

# D

rop

s

D = 1.1 mm

0 1 2 3 40

2000

4000

r (m)

# D

rop

s

D = 1.2 mm

0 1 2 3 40

2000

4000

r (m)

# D

rop

s

D = 1.4 mm

0 1 2 3 40

1000

2000

r (m)

# D

rop

s

D = 1.6 mm

0 1 2 3 40

500

1000

r (m)#

Dro

ps

D = 1.9 mm

0 1 2 3 40

500

r (m)

# D

rop

s

D = 2.1 mm

0 1 2 3 40

200

400

r (m)

# D

rop

s

D = 2.4 mm

0 1 2 3 40

200

400

r (m)

# D

rop

s

D = 2.8 mm

0 1 2 3 40

50

100

r (m)

# D

rop

s

D = 3.3 mm

0 1 2 3 40

5

10

r (m)

# D

rop

s

D = 3.8 mm

212

BETE WL-1 1/2

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.31 mm

0 1 2 3 40

5

10x 10

4

r (m)

# D

rop

s

D = 0.44 mm

0 1 2 3 40

5

10x 10

4

r (m)

# D

rop

s

D = 0.56 mm

0 1 2 3 40

5x 10

4

r (m)

# D

rop

s

D = 0.69 mm

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.81 mm

0 1 2 3 40

5000

10000

r (m)

# D

rop

sD = 0.94 mm

0 1 2 3 40

2000

4000

r (m)

# D

rop

s

D = 1.1 mm

0 1 2 3 40

1000

2000

r (m)

# D

rop

s

D = 1.2 mm

0 1 2 3 40

1000

2000

r (m)

# D

rop

s

D = 1.4 mm

0 1 2 3 40

500

1000

r (m)

# D

rop

s

D = 1.6 mm

0 1 2 3 40

200

400

r (m)#

Dro

ps

D = 1.9 mm

0 1 2 3 40

100

200

r (m)

# D

rop

s

D = 2.1 mm

0 1 2 3 40

100

200

r (m)

# D

rop

s

D = 2.4 mm

0 1 2 3 40

20

40

r (m)

# D

rop

s

D = 2.8 mm

0 1 2 3 40

5

10

r (m)

# D

rop

s

D = 3.3 mm

0 1 2 3 40

0.5

1

r (m)

# D

rop

s

D = 3.8 mm

213

BETE WL-3

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.31 mm

0 1 2 3 40

5

10x 10

4

r (m)

# D

rop

s

D = 0.44 mm

0 1 2 3 40

5

10x 10

4

r (m)

# D

rop

s

D = 0.56 mm

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.69 mm

0 1 2 3 40

1

2x 10

4

r (m)

# D

rop

s

D = 0.81 mm

0 1 2 3 40

5000

10000

r (m)

# D

rop

s

D = 0.94 mm

0 1 2 3 40

5000

r (m)

# D

rop

s

D = 1.1 mm

0 1 2 3 40

2000

4000

r (m)

# D

rop

s

D = 1.2 mm

0 1 2 3 40

2000

4000

r (m)

# D

rop

s

D = 1.4 mm

0 1 2 3 40

1000

2000

r (m)

# D

rop

s

D = 1.6 mm

0 1 2 3 40

1000

2000

r (m)#

Dro

ps

D = 1.9 mm

0 1 2 3 40

500

1000

r (m)

# D

rop

s

D = 2.1 mm

0 1 2 3 40

200

400

r (m)

# D

rop

s

D = 2.4 mm

0 1 2 3 40

500

r (m)

# D

rop

s

D = 2.8 mm

0 1 2 3 40

100

200

r (m)

# D

rop

sD = 3.3 mm

0 1 2 3 40

20

40

r (m)

# D

rop

s

D = 3.8 mm

214

Lechler 608

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.31 mm

0 1 2 3 40

5x 10

4

r (m)

# D

rop

s

D = 0.44 mm

0 1 2 3 40

5x 10

4

r (m)

# D

rop

s

D = 0.56 mm

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.69 mm

0 1 2 3 40

1

2x 10

4

r (m)

# D

rop

s

D = 0.81 mm

0 1 2 3 40

5000

10000

r (m)

# D

rop

s

D = 0.94 mm

0 1 2 3 40

2000

4000

r (m)

# D

rop

s

D = 1.1 mm

0 1 2 3 40

1000

2000

r (m)

# D

rop

s

D = 1.2 mm

0 1 2 3 40

500

1000

r (m)

# D

rop

s

D = 1.4 mm

0 1 2 3 40

500

r (m)

# D

rop

s

D = 1.6 mm

0 1 2 3 40

100

200

r (m)#

Dro

ps

D = 1.9 mm

0 1 2 3 40

20

40

r (m)

# D

rop

s

D = 2.1 mm

0 1 2 3 40

10

20

r (m)

# D

rop

s

D = 2.4 mm

0 1 2 3 40

5

10

r (m)

# D

rop

s

D = 2.8 mm

0 1 2 3 40

0.5

1

r (m)

# D

rop

sD = 3.3 mm

0 1 2 3 4-1

0

1

r (m)

# D

rop

s

D = 3.8 mm

215

Lechler 648

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.31 mm

0 1 2 3 40

5

10x 10

4

r (m)

# D

rop

s

D = 0.44 mm

0 1 2 3 40

5

10x 10

4

r (m)

# D

rop

s

D = 0.56 mm

0 1 2 3 40

5x 10

4

r (m)

# D

rop

s

D = 0.69 mm

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.81 mm

0 1 2 3 40

1

2x 10

4

r (m)

# D

rop

s

D = 0.94 mm

0 1 2 3 40

5000

r (m)

# D

rop

s

D = 1.1 mm

0 1 2 3 40

1000

2000

r (m)

# D

rop

s

D = 1.2 mm

0 1 2 3 40

1000

2000

r (m)

# D

rop

s

D = 1.4 mm

0 1 2 3 40

500

1000

r (m)

# D

rop

s

D = 1.6 mm

0 1 2 3 40

200

400

r (m)#

Dro

ps

D = 1.9 mm

0 1 2 3 40

100

200

r (m)

# D

rop

s

D = 2.1 mm

0 1 2 3 40

50

100

r (m)

# D

rop

s

D = 2.4 mm

0 1 2 3 40

50

r (m)

# D

rop

s

D = 2.8 mm

0 1 2 3 40

10

20

r (m)

# D

rop

sD = 3.3 mm

0 1 2 3 40

2

4

r (m)

# D

rop

s

D = 3.8 mm

216

Spray Systems 1/8GG

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.31 mm

0 1 2 3 40

5

10x 10

4

r (m)

# D

rop

s

D = 0.44 mm

0 1 2 3 40

5x 10

4

r (m)

# D

rop

s

D = 0.56 mm

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.69 mm

0 1 2 3 40

1

2x 10

4

r (m)

# D

rop

s

D = 0.81 mm

0 1 2 3 40

5000

10000

r (m)

# D

rop

s

D = 0.94 mm

0 1 2 3 40

2000

4000

r (m)

# D

rop

s

D = 1.1 mm

0 1 2 3 40

2000

4000

r (m)

# D

rop

s

D = 1.2 mm

0 1 2 3 40

2000

4000

r (m)

# D

rop

s

D = 1.4 mm

0 1 2 3 40

500

1000

r (m)

# D

rop

s

D = 1.6 mm

0 1 2 3 40

200

400

r (m)#

Dro

ps

D = 1.9 mm

0 1 2 3 40

100

200

r (m)

# D

rop

s

D = 2.1 mm

0 1 2 3 40

20

40

r (m)

# D

rop

s

D = 2.4 mm

0 1 2 3 40

10

20

r (m)

# D

rop

s

D = 2.8 mm

0 1 2 3 40

1

2

r (m)

# D

rop

sD = 3.3 mm

0 1 2 3 4-1

0

1

r (m)

# D

rop

s

D = 3.8 mm

217

Steinen 100

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

sD = 0.31 mm

0 1 2 3 40

5

10x 10

4

r (m)

# D

rop

s

D = 0.44 mm

0 1 2 3 40

5

10x 10

4

r (m)

# D

rop

s

D = 0.56 mm

0 1 2 3 40

5x 10

4

r (m)

# D

rop

s

D = 0.69 mm

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.81 mm

0 1 2 3 40

5000

10000

r (m)

# D

rop

s

D = 0.94 mm

0 1 2 3 40

2000

4000

r (m)

# D

rop

s

D = 1.1 mm

0 1 2 3 40

1000

2000

r (m)

# D

rop

s

D = 1.2 mm

0 1 2 3 40

1000

2000

r (m)

# D

rop

s

D = 1.4 mm

0 1 2 3 40

500

1000

r (m)

# D

rop

s

D = 1.6 mm

0 1 2 3 40

200

400

r (m)

# D

rop

s

D = 1.9 mm

0 1 2 3 40

50

100

r (m)

# D

rop

s

D = 2.1 mm

0 1 2 3 40

20

40

r (m)

# D

rop

s

D = 2.4 mm

0 1 2 3 40

5

10

r (m)

# D

rop

s

D = 2.8 mm

0 1 2 3 40

0.5

1

r (m)

# D

rop

sD = 3.3 mm

0 1 2 3 4-1

0

1

r (m)

# D

rop

s

D = 3.8 mm

218

Steinen 150

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

sD = 0.31 mm

0 1 2 3 40

5

10x 10

4

r (m)

# D

rop

s

D = 0.44 mm

0 1 2 3 40

5x 10

4

r (m)

# D

rop

s

D = 0.56 mm

0 1 2 3 40

2

4x 10

4

r (m)

# D

rop

s

D = 0.69 mm

0 1 2 3 40

1

2x 10

4

r (m)

# D

rop

s

D = 0.81 mm

0 1 2 3 40

5000

10000

r (m)

# D

rop

s

D = 0.94 mm

0 1 2 3 40

5000

r (m)

# D

rop

s

D = 1.1 mm

0 1 2 3 40

2000

4000

r (m)

# D

rop

s

D = 1.2 mm

0 1 2 3 40

2000

4000

r (m)

# D

rop

s

D = 1.4 mm

0 1 2 3 40

500

1000

r (m)

# D

rop

s

D = 1.6 mm

0 1 2 3 40

200

400

r (m)

# D

rop

s

D = 1.9 mm

0 1 2 3 40

100

200

r (m)

# D

rop

s

D = 2.1 mm

0 1 2 3 40

50

r (m)

# D

rop

s

D = 2.4 mm

0 1 2 3 40

20

40

r (m)

# D

rop

s

D = 2.8 mm

0 1 2 3 40

2

4

r (m)

# D

rop

s

D = 3.3 mm

0 1 2 3 40

0.5

1

r (m)

# D

rop

s

D = 3.8 mm

219

Measured Nozzle Raindrop Size Distribution

0 1 2 3 4 510

-4

10-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n, N

D (

mm

-1 m

-3)

BETE WL1

BETE WL1

Willis and Tattelman (1989)

Best (1950)

Marshall and Palmer (1948)

0 1 2 3 4 510

-4

10-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n, N

D (

mm

-1 m

-3)

BETE WL-1 1/2

BETE WL-1 1/2


Best (1950)


220

0 1 2 3 4 510

-4

10-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n, N

D (

mm

-1 m

-3)

BETE WL2

BETE WL2


Best (1950)


0 1 2 3 4 510

-4

10-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n, N

D (

mm

-1 m

-3)

BETE WL3

BETE WL3


Best (1950)


221

0 1 2 3 4 510

-4

10-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n, N

D (

mm

-1 m

-3)

Lechler 608

Lechler 608


Best (1950)


0 1 2 3 4 510

-4

10-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n, N

D (

mm

-1 m

-3)

Lechler 648

Lechler 648


Best (1950)


222

0 1 2 3 4 510

-4

10-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n, N

D (

mm

-1 m

-3)

Spray Systems 8W 1/8GG

Spray Systems 8W 1/8GG


Best (1950)


0 1 2 3 4 510

-4

10-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n, N

D (

mm

-1 m

-3)

Steinen 100

Steinen 100


Best (1950)


223

0 1 2 3 4 510

-4

10-2

100

102

104

106

Diameter (mm)

Co

nce

ntr

atio

n, N

D (

mm

-1 m

-3)

Steinen 150

Steinen 150


Best (1950)


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, 1993. 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 Penetration 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 relation. J. Geophys. Res. 104:6155–

6169.

Balderrama, JA, FJ Masters, KR Gurley, DO Prevatt, LD Aponte-Bermúdez, TA

Reinhold, JP Pinelli, CS Subramanian, SD Schiff, and AG Chowdhury, 2011. The Florida Coastal Monitoring Program (FCMP): A review. Journal of Wind 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 Distribution of Raindrops. Quarterly Journal Royal Meteorological Society 76(327):16-36.

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):441-462.

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 Measurements at Test Building Sites. The 7th symposium on Building Physics in the Nordic Countries, Reykjavík, 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. Building 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 Façades. Atmospheric


Bomberg, MT, MZ Rousseau, G Desmarais, M Nicholls, 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 bricks 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 Polarimetric Radar. Journal of Atmospheric and Oceanic Technology 1973 - 1993.

226

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 Atmospheric 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. Measurement 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 flat 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.

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.

Grossklaus, 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.

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 Research 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 assessment 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 Observation. 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).

Insurance Institute for Business & Home Safety 2009. HURRICANE IKE: Nature’s Force vs. Structural Strength. www.disastersafety.org/resource/resmgr/pdfs/hurricane_ike.pdf.

ISO, BS E. 2009. Hygrothermal performance of buildings – calculation and presentation of climatic 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 clustering 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 17:1057–1063.

Joss, J and A Waldvogel, 1967. A raindrop spectrometer with automatic readout. Pure Applied Geophysics 68:240–246.

229

Joss, J and A Waldvogel, 1970. A method to improve the accuracy of radar measured amounts of precipitation. Preprints, 14th Conf. on Radar 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 Short, 1994. A probability distribution model for rain rate. Journal of Applied Meteorology 33:1486-1493.

Lacasse, MA, TJ O’Connor, S Nunes, and P Beaulieu 2003. Report from Task 6 of 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 Tropical 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.

230

Lopez, CR, FJ Masters, and SR Bolton, 2011. Water Penetration Resistance of Residential 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. Westford,

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 Balderrama, 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 from 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 the Extreme Rainfall During Hurricane Danny’s (1997) Landfall. Monthly Weather Review 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 Martinez 2006. Water Intrusion in Central Florida Homes During Hurricane Jeanne in September 2004. Orlando, FL.

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 Measurement 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 Memorandum 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


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 wind 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 Rotation 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 Industrial

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.

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.

Rodgers, 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 and 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 Profiler Spectra using Deconvolution and a Nonlinear Least Squares Fitting Technique. Journal of Atmospheric and Oceanic Technology 19:964-874.

Schick, RJ 1997. An Engineer’s Practical Guide to Drop Size. Wheaton, IL.

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 Thunderstorms. 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 Observational 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).

Eindhoven

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 Rates. 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 Environment 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 Measurement 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 Meteorology 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, 1999. 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-Conditioning Engineers Thermal Performance of the Exterior Envelopes of Buildings 7:469-483.

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 using 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):1048-1058.

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.

235

BIOGRAPHICAL SKETCH

Carlos 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

business administrator and architect/contractor he was exposed to the construction

environment 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

behavior 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 companies 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 that 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 Florida, 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 Florida.

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.

To my parents, Amanda Duarte and Alfonso...

Documents

Transcript of To my parents, Amanda Duarte and Alfonso...