Structures 2008: Crossing Borders

© 2008 ASCE

Investigation of Thunderstorm Winds on an Instrumented Building Authors: Franklin T. Lombardo, Texas Tech University, Lubbock, TX, 79409, [email protected] Douglas A. Smith, Texas Tech University, Lubbock, TX, 79409, [email protected]

ABSTRACT The Texas Tech Wind Engineering Field Research Laboratory (WERFL) building is equipped with over 200 sensors to measure the pressure on the building. Since its inception, many pressure time histories have been documented. However, a full length thunderstorm record has never been fully studied for its effects on the building and its comparisons with larger scale synoptic winds. On August 24, 2007 a moderate thunderstorm downdraft occurred at the WERFL building with a peak gust approaching 50 mph. An investigation of this thunderstorm is contained within this paper. The large increase in wind speed and air density/pressure accompanied by downburst passage may cause increased pressures on the building. Also, the turbulent and thermodynamic characteristics of thunderstorm winds contribute to large vertical (upward) velocities which may in turn lead to stronger uplift pressures. These pressures on the building may lead to increased damage probabilities on low-rise structures. Comparisons with a non-thunderstorm record at a similar angle of attack are also carried out.

INTRODUCTION Winds produced from thunderstorms are typically generated when air is lifted in a thunderstorm updraft and therefore cools and condenses. Owing to negative buoyancy, this chilled or water loaded air is then forced downward where it comes in contact with the ground and spreads out in all directions due to ground effects and a pressure gradient force [1]. This creates a boundary between warmer environmental air and the air generated from the thunderstorm, typically called a gust front or downdraft [2]. Thunderstorm winds are highly unpredictable, complex and stochastic. The strong horizontal and vertical wind shears that accompany the thunderstorm downdraft [3] are threats to aircraft, launch vehicles and buildings. Thunderstorm downdrafts are also well known for their baroclinic (changes in density on a constant pressure surface) zones or density currents (horizontal vortex, with induced convective upward motion) [4] at their leading edge shown in Figure 1. This upward motion in conjunction with the horizontal vortex production have been known to create intense vertical vortices, known as gustnadoes, in addition to serving as a focus for additional severe weather [1].

Structures 2008: Crossing Borders

© 2008 ASCE

FIGURE 1 – THUNDERSTORM DOWNDRAFT STRUCTURE [5] Thunderstorm winds have also been shown to be very turbulent, fully three dimensional, non-linear and non-stationary. For this reason it makes it extremely difficult to characterize the statistical properties of thunderstorm winds for engineering analysis and these properties expected to be quite different from those of large scale winds [6]. Since these properties are expected to be different, it is important to analyze thunderstorms as a separate phenomenon. Strong thunderstorm winds are known to generate large economic losses [7] and dominate most temperate wind climates [8], making it imperative to consider their properties for structural design. In addition, maximum wind velocities are expected to occur in a thunderstorm downdraft typically occur at a height of 50 to 100 m or maybe even lower [9], which makes them a potential hazard to all structures that fall under the category of low-rise. A full scale investigation of the wind induced pressures from thunderstorms is not found in the literature, including that of the WERFL building. However, many thunderstorms were likely to have affected the site and a detailed analysis of past records correlated with thunderstorm records would likely contain multiple events. This downdraft will attempt to be explained in both physical and engineering aspects in the context of this paper, as well as identifying areas of ongoing and future research on this important topic.

THUNDERSTORM EVENT (AUGUST 24, 2007) The convectively generated winds on August 24 were due to the formation of a small Mesoscale Convective System [10]. Maximum wind speeds recorded from this event in areas in proximity to the Reese Center site were greater than 65 mph [11]. The initial thunderstorm downdraft in the vicinity of the WERFL building propagated well ahead of the main convective region and is shown in Figure 2 at ~ 3500 s. All wind speed and direction information were recorded by a 200 meter (656 feet) instrumented tower at levels of 13 and 33 feet in close proximity to the WERFL building. The WERFL building does employ a sonic anemometer that is strategically placed directly above the structure. However, estimates of vertical wind speeds were shown to be skewed from the probable mean (~ 0). This may be due to building effects or may represent some error with the anemometer itself. Therefore, data from the 200 meter tower was used to give a background, or estimated wind time series as it approached the WERFL building. As the convection approached, a second, stronger downdraft affected the WERFL building, with the maximum winds of approximately 50 mph occurring approximately at 6000s shown in Figure 2. Additional convection may have developed and caused additional non-stationary wind speeds between 8000s and 10000s. The wind direction shifted initially from south-southeasterly to a

Structures 2008: Crossing Borders

© 2008 ASCE

northeasterly component due to the pressure gradient from the oncoming downdraft. Later on in the time series the downdraft may exhibit a brief backside peak noted in many thunderstorm observations (~ 7200s) where the wind direction quickly shifts to a southeasterly direction and continues up to ambient conditions before being affected by weaker convection. A)

B)

FIGURE 2 – WIND SPEED AND DIRECTION TIME HISTORY FROM 200 METER TOWER AT A) 13 FEET AND B) 33 FEET (30 HZ)

BUILDING EFFECTS There have not been any documented occurrences of thunderstorm winds affecting a full-scale instrumented building in the literature, although at sites such as the WERFL building, thunderstorm flow likely affected the structure numerous times. There have been however, many computer simulations of both downdraft (thunderstorm) and tornado phenomena. Nicholls et al. [12] found in a computer simulation of thunderstorm winds on a small low-rise building, there were large negative pressure coefficients (Cp) at the leading edge of the roof and the flow showed no reattachment. Sarkar et al. [13] showed in simulation of both thunderstorm and tornadic winds that Cp’s were larger than those of synoptic boundary layer conditions and the uplift loads exceeded that of current design specifications. Letchford [14] has also shown large negative Cp’s in thunderstorm simulations and showed underestimation using quasi-steady assumptions. Selvam and Millett [15] performed a large eddy simulation of tornadic winds on a structure and found the large vertical wind component of the tornado when it strikes the modeled building becomes highly concentrated. Due to this, the force coefficients (Cf) in the vertical were found to be higher than that of traditional boundary layer winds, and found Cp’s approached -40 at the leading edge of the roof. In a later, refined simulation, Selvam and Millett [16] found that in tornadoes, the recirculation of vorticity produces highly localized suction in the roof, and that in high vertical velocity occurrences, flow separation occurs just above the entire roof surface. In addition, it was found that higher Cp values in tornadic winds than in boundary layer conditions. Although, these two studies [15, 16] looked at tornadic winds, the large vertical velocities and horizontal and vertical vorticity are also applicable to thunderstorm winds on a smaller scale. The presence of relatively large vertical velocities could delay or prevent separation [12], therefore increasing the area of strong uplift pressures over the roof of a structure. A typical thunderstorm downdraft diagram is shown in Figure 1.

Structures 2008: Crossing Borders

© 2008 ASCE

Secondly, the colder, denser air that usually accompanies thunderstorm downdraft may cause increased load on the structure due to the rapid increase in air density and atmospheric pressure. It was reported in [17] that this increase and other variations of pressure within the internal downdraft structure, may cause increased load in sealed structures. Along with this denser air comes the rapid increase of wind speed as shown in Figure 2. This increase of wind speed over ambient conditions would likely cause a subsequent rapid increase in load on the structure in question. Therefore at least three important characteristics in thunderstorms need to be further identified when considering their impacts on structures:   

Importance of vertical velocity component and associated vorticity (horizontal, vertical) Introduction of higher density/pressure air Initial increase in wind speed over ambient conditions and subsequent internal increases

In the August 24, 2007 event, the vertical velocities that accompanied the thunderstorm onset at heights of l3 and 33 feet are shown in Figure 3. This figure shows the highly turbulent nature of the event showing both large positive (upward) and negative (downward) vertical velocities (from about +10 to -10 mph) at 33 feet. Stronger vertical velocities near the surface have been recorded with other thunderstorm events at the Reese field site. It is fairly evident from this information to hypothesize that when there are large fluctuations in vertical velocity, it is likely some convective element producing these fluctuations and more specifically from the turbulent vortices that develop in the internal structure of the downdraft [18]. In addition, the initial uplifting of warm air before the onset of the downdraft (not shown) reached vertical velocities of about 6 (+) mph at the 33 feet height.

FIGURE 3 – VERTICAL VELOCITIES AT A) 13 FEET B) 33 FEET FROM 200 METER TOWER The main region of interest for this event from a structural standpoint was determined to be approximately between 3000 and 6000s. The wind speed time history is obviously nonstationary, but the mean wind speed ( U ) was determined to be 25 mph and the mean angle of attack (  ) on the WERFL building was determined to be 317 degrees. Due to the nonstationarity of the wind speed record and the normalization of Cp’s by the mean wind speed

Structures 2008: Crossing Borders

© 2008 ASCE

over the course of the record, the measured Cp’ s from this record were determined to be erroneous. However with information of temperature, barometric pressure, and relative humidity, an air density time history can be calculated and along the mean wind speed the calculation of an actual pressure time history is possible. This information is all available at 13 feet and in particular the increases in pressure and density are shown in Figure 4. There were numerous other heights that measured wind speed, but do not measure barometric pressure. Air density along with the mean wind speed can be used to determine pressures using quasi-steady assumptions. 26.6 80

70

1.09 26.55

70

65

60 60

50

55 0

40 5000

10000

0

1.08

26.5

1.07 1.06

26.45 5000

10000 0 Time (s)

5000

10000

1.05 0

5000

10000

FIGURE 4 – TIME HISTORIES OF A) TEMPERATURE (F), B) RELATIVE HUMIDITY (%), C) BAROMETRIC PRESSURE (IN OF HG) AND D) AIR DENSITY (KG/M3) FROM 13 FEET. THE SOLID YELLOW LINE IN D) IS A 30 SECOND AVERAGE AIR DENSITY.

Non-Thunderstorm Record For future comparison, a non-thunderstorm record (15 minute) at similar mean wind speed and angle of attack as compared to the thunderstorm event was retrieved from the WERFL archives and will be analyzed to determine the differences, if any, between the two wind types. The mean wind speed ( U ) was 24 mph while the mean wind direction, which over the course of the thunderstorm exhibited relatively small changes in direction, was clockwise from true north (  ) at approximately 317 degrees, making the mean angle of attack as measured by WERFL (  ) 43 degrees, simply located at a reflected tap from the thunderstorm case. Taps 24008 (windward wall), 53030 (leading edge of roof), 52020 (roof) and 50515 (roof) were analyzed in this case in an attempt to stay symmetrical in regards to the thunderstorm case and are shown in Figure 5. These locations and other locations throughout the WERFL building will need to be studied further in comparison to thunderstorm winds.

Structures 2008: Crossing Borders

2

A)

© 2008 ASCE

0 B)

C)

-0.5

1.5 1

-0.4 -1

Cp

-2 -2.5

0

-0.6 -0.8

-1.5

-3

-0.5

-1

-3.5 -1

-2 -1.2

-4 0

200 400 600 800

-4.5

D)

-0.2

-1.5

0.5

0

-0.5

-1

-1.5

0

0

-2.5 200 400 600 800 0 Time (s)

200 400 600 800

-1.4

0

200 400 600 800

FIGURE 5 – CP TIME HISTORY (NON-THUNDERSTORM) FOR TAPS A) 24008 B) 53030 C) 52020 AND D) 50515

CONCLUSIONS AND FUTURE STUDY A moderate thunderstorm affected the Reese Field Site including the WERFL building on August 24, 2007, producing winds approaching 50 mph, and lasting for approximately 3000s. All significant increases in wind speed were accompanied by increases in the vertical velocity component. This component may increase (decrease) instantaneous values of Cp over those of their non-thunderstorm counterparts, depending on the pressure tap location. Although the WERFL building was not operational at periods during the 2007 thunderstorm season, it will be fully operational for the beginning of the 2008 season. In addition, many thunderstorm events were recorded using the nearby 200 meter tower in both 2006 and 2007 as well as the sonic anemometer mounted above the WERFL building. These events can be used to generate simulated pressure time histories using quasi-steady methods. The quasi-steady method does not carry the assumption of stationarity, making it valid for thunderstorm events. The predicted time histories of pressure will be compared with the measured time histories of pressure in a thunderstorm, and a residual time history (difference between the two time histories) can be created and compared with non-thunderstorm winds. This information may eventually be used to compare with the mean Cp’s prescribed in the structural design standard [19]. In some cases, and mostly in the strongest cases recorded at the Reese Field Site, multiple boundaries or surges within the internal structure of the downdraft have been prominent. Other studies [4, 20] have noted this phenomenon and an example from July 2007 from Reese Center that displays cyclical boundaries is shown in Figure 6. The event discussed in this paper may have had multiple boundaries but their presence was not as evident as in some other cases. This has been largely unaccounted for in wind tunnel simulations and may warrant further study due to the rapid loading variations on the structure due to the wind speed itself, vertical velocity gradients and rapid pressure changes.

Structures 2008: Crossing Borders

© 2008 ASCE

60

Wind Speed (mph)

50 40 30 20 10 1400

1450

1500 1550 Time (s)

1600

1650

FIGURE 6 – THUNDERSTORM SHOWING MULTIPLE BOUNDARIES (SURGES) – 200 METER TOWER, JULY 10, 2007.

ACKNOWLEDGEMENT The first author would like to thank the National Science Foundation (NSF) and specifically NSF Grant No. 0221688 for funding a portion of this research.

REFERENCES [1] Fujita, T. (1985). The Downburst, SMRP 210, U. of Chicago Press [2] Markowski, P., Hannon, C., Rasumussen, E. (2006). “Observations of convective initiation “failure” from the 12 June 2002 IHOP deployment”, Mon. Wea. Rev., 134, 1, 375-405 [3] Sinclair, R.W., Anthes, R.A., Panofsky, H.A. (1973). “Variation of the Low Level Winds During the Passage of a Thunderstorm Gust Front”, NASA Contractor Report, NASA CR-2289, 65 pp. [4] Goff, R.C. (1976). “Vertical Structure of Thunderstorm Outflows”, Monthly Wea. Rev., 104, 1429-1440 [5] Martner, B.E. (1997). “Vertical Velocities in a Thunderstorm Gust Front and Outflow”, J. Applied Meteorology, 36, 615-622 [6] Choi, E.C.C. (2004). “Field measurement and experimental study of wind speed profiles during thunderstorms”, J. Wind Eng. Ind. Aerodyn., 92, 275-290 [7] Crandell, J.H., Farkas, W., Lyons, J.M. (2000). “Near-ground wind and its characterization for engineering applications”, Wind and Structures, 3, 3, 143-158. [8] Holmes, J.D. (2001). Wind Loading on Structures. Spon Press, London-NY [9] Kim, J. and Hangan, H. (2007). “Numerical simulations of impinging jets with application to downbursts”, J. Wind Eng. Ind. Aerodyn., 95, 4, 279-298 [10] Wakimoto, R. (2001). “Convectively Driven High Wind Events”, Met. Monographs, Severe Convective Storms, Dowsell, C. ed. [11] Texas Tech University (2007). West Texas Mesonet. www.mesonet.ttu.edu (Accessed December 6, 2007) [12] Nicholls, M., Pielke, R., Meroney, R. (1993). “Large eddy simulation of microburst winds flowing around a building”, J. Wind. Eng. Ind. Aerodyn., 46-47, 229-237 [13] Sarkar, P.P., Haan Jr., F.L., Balaramudu, V., Sengupta, A. (2006). “Laboratory Simulation of Tornado and Microburst to assess Wind Loads on Buildings”, Structures Congress 2006, St. Louis, MO, CD-ROM. [14] Letchford, C.W. (2002). “Thunderstorms – their importance in wind engineering (a case for the next generation wind tunnel)”, J. Wind Eng. Ind. Aerodyn., 90, 1415-1433 [15] Selvam, R.P. and Millett, P.C. (2003). “Computer modeling of tornado forces on buildings”, Wind and Structures, 6, 3, 209-220 [16] Selvam, R.P. and Millett, P.C. (2005). “Large eddy simulation of the tornado-structure interaction to determine structural loadings”, Wind and Structures, 8, 1, 49-60

Structures 2008: Crossing Borders

© 2008 ASCE

[17] Chay, M.T. and Letchford, C.W. (2002). “Pressure distributions on a cube in a simulated thunderstorm downburst – Part A: stationary downburst observations”, J. Wind Eng. Ind. Aerodyn., 90, 711-732 [18] Järvi, L., Punkka, A.J., Schultz, D.M., Petäjä, T., Hohti, H. et al. (2007). “Micrometeorological observations of a microburst in Southern Finland”, Boundary-Layer Meteorol, 125, 343-359 [19] American Society of Civil Engineers (ASCE). (2006). Minimum Design Loads for Buildings and Other Structures, ASCE/SEI 7-05, Reston, VA. [20] Klingle, D.L. (1987). “An investigation of the internal structure of the thunderstorm outflow with particular attention to multiple surges”, Ph.D. Thesis, Purdue University