[ ]Preliminary Results of Full-Scale Monitoring of Hurricane Wind Speeds and

Wind Loads on Residential Buildings

Peter L. DatinGraduate Research

Assistant

David O. PrevattDirector and

Assistant Professor

Wind Load Test Facility

Department of Civil Engineering

Clemson University

[ ]Outline

• Introduction• The Florida Coastal Monitoring Program (FCMP)• Goals of the FCMP• In-field experimental methods

– Mobile tower system– Instrumented houses– Public access to collected data

• Wind tunnel testing• Preliminary results – comparison of full-scale to

wind tunnel• Summary and preliminary observations• Future work

[ ]Introduction

• Hurricanes continue to cause severe damage to residential structures

• Historically, these tests derived from frontal weather systems

• Need to validate wind tunnel results for extreme wind events

• Wind tunnel tests can only provide limited knowledge of wind loads on full-scale structures

• Necessary to improve our understanding of the wind-structure interaction during extreme wind events

[ ]• The Florida Coastal

Monitoring Program was started in 1998

• Research venture between– Clemson University– University of Florida– Florida International

University– Institute for Business and

Home Safety• Sponsored by:

– Florida Department of Community Affairs

– SC and FL Sea Grant Consortia

• Objectives– In-field measurement of

hurricane wind velocities and wind-induced pressures on residential buildings

– Wind tunnel studies to compare with full-scale data

• FCMP mobilizes before a hurricane makes landfall placing instrumentation in the path of the storm

[ ]Portable Hurricane Instrumentation

• 32 homes in Florida and 6 in the Carolinas are pre-wired to be instrumented

• Absolute pressure transducers record pressures at critical locations on the roof

• Establish a reference pressure to measure atmospheric pressure that must be subtracted from the absolute pressure to obtain actual wind pressures

• Stiff 10-meter towers placed in hurricane’s path

• Measure wind velocities, temperature, barometric pressure, etc.

Pressure Sensors

Computer Box

[ ]Public Access to Collected Data

• High resolution wind speed data available– 10 Hz sampling rate– 15-minute mean wind speed– 3-second gust wind speed

• Available in near real-time on the FCMP website

• Used in official NOAA tropical cyclone reports

FCMP Websitewww.ce.ufl.edu/~fcmp

[ ]Measured Hurricane Ivan Wind Speeds

Measured wind speeds during Hurricane Ivan (2004) converted to 3-second gust at 10-meters

[ ]Atmospheric Boundary Layer Wind Tunnel

• Atmospheric boundary layer wind tunnel

• Open circuit wind tunnel• 3-meters wide by 2-

meters high

• Total length – 100 feet• Can simulate various

terrains• Model sizes from 1:50 to

1:500

[ ]Pressure Coefficients

• Non-dimensional quantification of wind pressures

• Can derive pressure coefficients from full-scale and wind tunnel

• Derive pressure coefficients based on 3-sec gust wind speed

• Allows direct comparison between full-scale and wind tunnel values

• Allows direct comparison with building code provisions

2

0

20

0

11

2

p

P P UC

UU

[ ]Example: FL-27 (GBB)

[ ]FL-27 Measured Wind Speeds

Tropical Storm Isidore (2002) Hurricane Ivan (2004)

[ ]FL-27 Sensor Layout and Model

Tropical StormIsidore Peak

Wind Direction

Hurricane Ivan Peak

Wind Direction

[ ]

Tropical StormIsidore Peak

Wind Direction

Hurricane Ivan Peak

Wind DirectionASCE 7-02GCp,min= -2.60

-1.09

-1.17

-3.01

-0.83

-0.71

-0.90

-3.00

-0.71

N/A

-0.57

-2.53

-0.55

-0.87

-1.39

-2.83

-1.10

-0.70

-0.55

-2.52

-0.48

-0.77

-0.52

-2.62

-0.85

-0.69

-0.58

-3.29

-0.59

-0.80

-0.63

-2.72

-0.50

-0.98

-0.52

-2.94

-0.57

-1.02

-0.80

-2.80

-2.63

-1.33

-2.00

-3.38

-2.24

-0.98

-0.64

-2.34

-1.15

-1.50

-2.00

-3.43

-1.46

-1.21

-0.81

-2.52

-0.62

-0.96

-0.71

-3.53

-0.71 -1.06

-1.08

-3.53

-1.21

-1.25

-1.16

-2.75

-1.32

-2.68

-1.64

-3.75

-1.44

-2.29

-1.66

-3.88

-1.39

-1.55

-0.80

-3.11

-1.32

Isidore FS

Isidore WT

Ivan FS

Ivan WT

Preliminary Results – Peak Minimum Cp Values

[ ]

y = 1.0542x

R2 = 0.2097

0.00

0.25

0 0.05 0.1 0.15 0.2 0.25

Model

Fu

ll-S

ca

le

y = x

(135 degrees)

Model

0.25

Ful

l-Sca

le

0.250.20.150.10.0500.00

y = 1.0542xR2 = 0.2097

y = x

RMS of Pressure Coefficients

Full-Scale vs. Wind Tunnel

-1.00

0.00

-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0

Model

Fu

ll-S

ca

le

y = x

(135 degrees)

y = x - 0.5

Model

Ful

l-Sca

le

-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0

1.0

y = x

y = x – 0.5

Mean Pressure Coefficient Values

RMS of Pressure Coefficients

y = 1.0542x

R2 = 0.2097

0.00

0.25

0 0.05 0.1 0.15 0.2 0.25

Model

Fu

ll-S

ca

le

y = x

(135 degrees)

Model

0.25

Ful

l-Sca

le

0.250.20.150.10.0500.00

y = 1.0542xR2 = 0.2097

y = x

-6.00

-5.00

-4.00

-3.00

-2.00

-1.00

0.00

-3 -2.5 -2 -1.5 -1 -0.5 0

Model

Fu

ll-S

cal

e

y = x

(135 degrees)

y = x - 1.0Ful

l-Sca

le

Model

0-1.5 -1.0 -0.5

-1.0

-2.0

-3.0

-4.0

-5.0

-6.0

y = x

y = x – 1.0

-3.0 -2.5 -2.0

Peak Minimum Pressure Coefficients

• Continuing research to determine sources of error

• Possible sources of error:

– Inaccurate regression analysis

– Wind tunnel models do not accurately simulate turbulence and wind speed for suburban terrain

– Limitations of instrumentation accuracy

• Adjustments to data:

– 0.5Cp corresponds to 2 psf in full-scale

– 20% increase in model Cp based on a different estimate of terrain roughness

– By applying these changes there is close agreement between model and full-scale values

[ ]Adjusted Results

-3

-2

-1

0

1

0 5 10 15 20 25 Pressure Tap Location

Cp Values

Model Cp Mean Model Cp Min Model Cp Max FS Cp Mean FS Cp Min Average Cp Max:

Cp

Val

ues

Pressure Tap Location2520151050

-3

-2

-1

0

1

[ ]Summary and Preliminary Observations

• Meteorological data used in civil engineering applications to provide greater understanding of wind characteristics and interaction with structures

• Unique data set on common residential building shapes subjected to hurricane force winds

• Linear regression shows agreement between full-scale and model scale, but this may not accurately represent the data distribution

• Loads measured at full-scale may not represent the worst wind loading condition since it is only from one wind direction

• First step in addressing continuing failures in components and cladding of residential buildings

[ ]Future Work

• Development of a reliability model of the data to provide a statistical basis for estimating the wind loads

• Further analysis of collected data from 5 additional houses in 2004 and another 5 in 2005

• Distribution of wind loads through the structure

• Continuing importance in collecting and making meteorological data available to the research community

• Future research results will be posted at:

davidoprevatt.com