1. Advisor
Dr. Dat Duthinh
Materials & Structural Division
Engineering Laboratory
Rahul Menon
Analyzing Wind Pressure on Building
Enclosures

2.  Background
 Project Description
 Contour Plots
 Introduction to Extreme Value Distributions
 Comparing Different Distributions
Outline

3. Hurricane Wilma

4.  Wind is chaotic and random
 Its speed fluctuates in time and space
 Need to design for peaks of pressure
Background

5.  Different building scale models were tested
 Variables Changed:
 Eave height
 Model Size
 Roof Slope
 Width
 Length
 Terrain Type
 cp: unit less coefficient describing wind pressure
Wind Tunnel Testing

6. Fluctuating Wind Pressure

7.  Plot 2D and 3D contour maps of pressure coefficients at
unevenly distributed taps
 Analyzes the peak pressure coefficients and wind direction
of chosen tap
 Alternative method of estimating peak wind pressure with
a given probability of exceedance
Software

8. Low-Rise Gable Roofed Buildings
𝛼
s
1
3
2

9. Time Series Data

10. 3D Contour Maps - 90º
Face 3Face 2

11. 3D Contour Maps - 315º
Face 3Face 2

12. What Do We Design For?

13.  Fit peak values to a distribution function
 Able to estimate values with a probability of exceedance
 Only limit distribution for large sets of random variables
Introduction to Extreme Value Distributions

14.  Unbiased
 The estimator function, which is a random variable, should equal the
value being estimated
𝐸 𝑓 = 𝑙𝑜𝑐𝑎𝑡𝑖𝑜𝑛 𝐸 𝑔 = 𝑠𝑐𝑎𝑙𝑒 𝐸 ℎ = 𝑠ℎ𝑎𝑝𝑒
 Minimum Variance
 The estimator function becomes more accurate as sample size
increases
Conditions for Extreme Value Distributions

15.  Gumbel (Type I):
𝐹 𝑥; 𝜇, 𝜎, 0 = 𝑒−𝑒
−
𝑥−𝜇
𝜎
, 𝜀 = 0
 Frechet (Type II):
𝐹 𝑥; 𝜇, 𝜎, 𝜀 = 𝑒−(
𝑥−𝜇
𝜎 )
−
1
𝜀
, 𝜀 > 0
 Reversed Weibull (Type III):
𝐹 𝑥; 𝜇, 𝜎, 𝜀 = 𝑒−(−
𝑥−𝜇
𝜎 )
1
𝜀
, 𝜀 < 0
Types of Extreme Value Distributions

16. Type I Distribution (BLUE method)
P = 0.78 Min
0 Partitions -2.2248
8 Partitions -3.0747

17. Type III Distribution (GEV)
P = 0.78 Min
Type III -1.1763
Type I -2.2248

18.  Made tools that will be used to analyze wind pressure data
from wind tunnels
 Extended peak analysis to more general extreme value
distributions
Conclusions

19.  Dr. Dat Duthinh
 Emil Simiu
 Adam Pintar
 Joe Main
 Luke Amatucci
 Matt Kovarek
 SURF Directors
 SURF Peers
Acknowledgements

20.  http://fris2.nist.gov/winddata/uwo-data/uwo-data.html
 http://fris2.nist.gov/winddata/uwo-data/blwt-ss20-2003.pdf
 http://www.itl.nist.gov/div898/winds/homepage.htm
 https://en.wikipedia.org/wiki/Generalized_extreme_value_
distribution
 http://www.palmbeachpost.com/news/weather/hurrican
es/hurricane-wilma-five-years-later-storm-taught-
hard/nMByM/
Citations