Throughout much of May 2024, a powerful heat dome settled over the Caribbean Sea and Gulf of Mexico, leading to scorching temperatures that shattered records across Central America. As the hot air at the dome’s northern boundary met cooler, drier air in the U.S., atmospheric instability gave rise to severe thunderstorms in the south-central United States. On May 14, the Storm Prediction Center (SPC) issued a level 2 severe weather alert for parts of Texas, which escalated to level 3 by the morning of May 16, extending toward the Gulf Coasts of Texas and Louisiana. In southeastern Texas and southwestern Louisiana, a northward-moving warm front heightened moisture levels, increasing the potential for storms to intensify along the boundary. By mid-afternoon, a significant mesoscale convective system developed across central and eastern Texas, generating numerous updrafts. As the system advanced southeast, it evolved into a derecho—an extensive, long-lived wind event—striking the Greater Houston area with winds up to 45 m/s and producing three EF1 tornadoes. By early May 17, the storm moved offshore into the Gulf of Mexico, reducing the severe weather threat inland but leaving behind significant damage along the Gulf Coast, including shattered windows, damaged roofs, and numerous downed trees and power lines (National Weather Service, 2024b; Voiland, 2024; Lewis, 2024). A map indicating the progression of the derecho through Houston area is shown in Figure 1. It is noteworthy that the buildings in question were not affected by the tornadoes. Figure 2 shows a map of the location of the studied buildings along with gust speeds recorded at various measurement points from ASOS observation network stations during the derecho (ASOS, 1998). The values indicate a range of 28 m/s to 32 m/s. However, as noted by the reconnaissance team (Kalliontzis et al., 2024), post damage survey reported higher wind speeds reaching 45 m/s in downtown Houston. For example, CoreLogic provided a map of wind speeds that showed values in the range of approximately 36 m/s to 40 m in the location of the buildings (INRIX, 2024).

Hurricane Beryl was a significant and devastating storm that struck the Caribbean, the Yucatán Peninsula, and the Gulf Coast of the United States in late June and early July 2024. On 8 July 2024, Hurricane Beryl impacted Houston as a Category 1 hurricane with gust speeds reaching approximately 40 m/s. Marking a historic event, it rapidly intensified becoming the earliest Category 5 hurricane ever recorded, reaching maximum sustained winds of 74 m/s. The hurricane originated from a tropical wave off the coast of Africa on 25 June 2024, and rapidly intensified, making its first landfall in Grenada as a high-end Category 4 storm on July 1. Beryl reached its peak strength as a Category 5 hurricane upon entering the Caribbean Sea, but soon weakened due to wind shear. The storm’s intensity fluctuated, briefly strengthening back to Category 3 before ultimately making its final landfall in Texas as a Category 1 hurricane on July 8th with sustained winds of 36 m/s. Beryl transitioned to a post-tropical cyclone over Arkansas on July 9 and eventually dissipated over Ontario on July 11 (National Weather Service, 2024a; KHOU 11 Staff, 2024). The track of the storm can be seen in Figure 3. Figure 4 shows a map of the location of the studied buildings along with gust speeds recorded at various measurement points from ASOS observation network stations during hurricane Beryl (ASOS, 1998). The values are showing very narrow range indicating that the wind speeds at the building locations are in this range from 37 m/s to 40 m/s.

This building, located in 1,500 Louisiana St, is a 40-story 182.9 m tall building. The building features a curtain wall façade where the horizontal structural beams support the vertical mullions to which the glass panels are fitted. Despite being a valuable option for construction of curved surfaces, the glass panels add to the lateral stiffness of the building which can cause glass failure in extreme events due to the material’s brittleness (Kalliontzis et al., 2024). Similar observations were also reported by (Alawode, et al., 2023) in full-scale testing of single-skin facades where the authors reported that increasing the stiffness of the joints by added vertical protrusions resulted in higher vibration and higher dynamic amplification factors. During the derecho event, the building’s cladding system was significantly damaged as shown in Figure 5. Figure 5A highlights the building’s location, identifying the damaged side and the direction of the derecho. The intense wind forces caused large and small façade panels to dislodge as shown in Figures 5B–D. Numerous glass windows shattered, raining debris onto the streets below, creating hazardous conditions in the surrounding areas. The damage was predominantly observed on the side facing another tall building, Chevron Corporation Tower, suggesting that the wind channeling effect between the adjacent towers may have amplified the negative pressures (suctions) on the facade. Interestingly, although the Chevron Corporation Tower shares the same oblong shape and orientation and is positioned in front of the Chevron Building Auditorium, it sustained minimal window damage. This discrepancy suggests that the significant damage to Chevron Building Auditorium was likely intensified by channeling effects, particularly as damage was concentrated at higher elevations, reducing the likelihood of wind-borne debris as the primary cause.

During Hurricane Beryl, the building sustained additional but minor impacts, according to company spokesperson Randy Stuart (Miranda, 2024). Figure 6 illustrates the relatively minor impacts observed during hurricane Beryl. It is notable that in both events, damage was concentrated on the curved surfaces of the oblong-shaped buildings, likely due to high suction from flow separation. Similar observation was reported for the same building during hurricane Ike which had wind gusts up to 41 m/s (Konz, 2009). Additionally, inaccurate manufacturing of the curved surface or errors in the installation can cause uneven distribution of the wind loading resulting in concentration of stresses at specific parts of the façade components possibly leading to failures. The building suffered less damage in Hurricane Ike compared to the derecho. The comparatively lower damage from Hurricanes Beryl and Ike, despite comparable gust speeds, underscores, in addition to the importance of interference effects, the need to further study the impact of the different loading mechanism associated with derechos and other related phenomena like thunderstorms as these events produce sudden, intense bursts of wind that can create highly localized stress concentrations on building façades.

This building, located in 1,111 Louisiana St, is a 53-story 226 m tall building. The building location and sustained damage is shown in Figure 7, with the damaged corner with respect to derecho’s path shown in Figure 7A. The building features a double-skin façade system supported by vertical mullions as shown in Figure 7D. The damage to this building during the derecho was mostly concentrated at one corner as can be seen in Figure 7B. This damage may be attributed to the cornering effects in rectangular buildings accompanied by the channeling effects caused by the two tall buildings facing that corner. Both façade layers were damaged in some cases mostly on the corners while some cases had damage to a single layer of the façade mostly away from the corner as depicted in Figure 7C. A detailed view of damage to both layers is shown in Figure 7D opposed to damage in the external layer only in Figure 7E. It is worth mentioning that the configuration of double skin facades, including the layout and the width of the air gap, affects the wind induced pressures and may contribute to the damage of the façade as some configurations have demonstrated higher pressures compared to those for single-skin facades (da Silva and Gomes, 2008; Lou et al., 2012). Therefore, it’s recommended to conduct wind tunnel testing for such type of façade to investigate all the design parameters. On the other hand, the building suffered very minor damage during the impact of hurricane Beryl.

This building, located in 1,001 Louisiana St, is a 35-story 153 m tall building. Detailed information about the cladding and building renovation, which was completed in 2012, can be found in (Abendroth L, 2013). The building suffered from considerable damage during the derecho, as shown in Figure 8A, where the damage is concentrated on the side of the building close to another tall building highlighting again the effect of channeling in intensifying the wind pressures. The direction of the derecho along with the damaged side is shown in Figure 8B. When combined with corner effects, most of the damage is concentrated on both corners of the damaged side, similar to CenterPoint Energy Plaza, as can be seen in Figure 8C. It is worth noting that the horizontal and vertical ribs might have affected the local pressure of building façade as these ribs might slightly increase the pressure on the building as noted by several researchers (Cheng et al., 2021). Additionally, the horizontal ribs can significantly affect the location of the stagnation point moving it downward (Liu et al., 2021). This finding aligns with the observation of damage concentrated more on the lower half of the building. However, this preliminary conclusion needs to be investigated more in light of the strong interference effects experienced by the building to study the complex interaction between multiple factors. Although the photos provided by (Kalliontzis et al., 2024) do not offer clear views of the building following hurricane Beryl, the few pictures, where the building was visible, are enough to conclude that the building had minimal damage from the hurricane.

This building, located in 1,000 Main St, is a 36-story 158 m tall building. The building sustained damage during the derecho as shown in Figure 9, with the building location with respect to the direction of both events shown in Figure 9A. On the lower half of building, less severe damage is observed as shown in Figures 9B, C which indicates only cracking of glass that may be attributed to wind-borne debris since the damage side is the windward wall. Severe damage was concentrated on the upper half of the building as shown in Figure 9D. On the lower half of building, less severe damage is observed as shown in Figure 9B which indicates only cracking of glass that may be attributed to wind-borne debris since the damage side is the windward wall. The same building experienced less damage on the same side during Hurricane Beryl as shown in Figure 10 where only minor cracking in the façade was observed.

This building, located in 1,415 Louisiana St, is a 44-story 168 m tall building. This building, also, suffered considerable damage in the derecho compared to the hurricane. Figure 11 shows the damage in the tower during the derecho indicating widespread breakage of the glass panels compared to minimum damage encountered during the hurricane as shown in Figure 12.

In summary, the damage to tall buildings during the Houston Derecho and Hurricane Beryl highlighted several critical lessons. The event underscored the vulnerability of glass facades in high-rise buildings, with significant damage observed, during the derecho with gust of 45 m/s, despite designed to withstand wind speeds up to 67 m/s as previously mentioned. This highlights the need to reevaluate current design and construction guidelines for façade elements, considering the unique loading characteristics associated with non-hurricane events, such as thunderstorm winds in derechos and downbursts. Additionally, the channeling effects of wind in dense urban areas may significantly alter the flow around tall buildings, further inducing more progressive damage. The abrupt change in wind speeds during transient events may have contributed to the observed damage in façade elements which again motivates the need to study the differences in loading between hurricanes and other transient events such as downbursts. These effects need to be studied in conjunction with the dynamic behavior of the façade system to better understand the interaction between the loading and the system as considered for Atmospheric boundary layer winds in Alawode et al. (2023), Bakhtiari et al. (2024). Additionally, it is recommended for reconnaissance teams to provide more data related to the design of facades to facilitate interpretation of the causes of damage. Similar recommendations were provided by Mayercsik and Bennett (2024), where the authors suggested incorporating facade directionality, and site-specific factors into forensic assessments to pinpoint the causes of failure and distinguish between one-time event and long-term effects. Key factors include solar exposure, that affect thermal expansion stresses, hail swaths and wind driven rain.

The building model is not intended to be a specific case from the damage observations presented in the previous section but rather a generic case that suits future testing considerations, such as aeroelastic testing which can be beneficial to understanding the effect of building vibration on the performance of the façade (Chen et al., 2023) and also to study the effect of transient events on tall building response (Kwon and Kareem, 2009). The building model is a 1:350 scale of a prototype measuring 48 m × 48 m × 347 m. With scaled dimensions of 13.7 cm × 13.7 cm × 99 cm as shown in Figure 14A, the model is equipped with 300 pressure taps, 75 taps on each surface, as depicted in Figure 14B, to measure the incident pressure on all surfaces at a sampling rate of 625 Hz using the SCANIVALVE system.

The isolated building will be tested under ABL and downburst for both 0° and 45° wind directions. Interference testing will consider 5 cases where a similar model is located upstream the base model with varying upstream distance. Figure 15 shows a schematic for the location of the interfering building with respect to the base model and wind direction for the five testing cases.

Figure 16 shows a typical comparison of the wind profiles and time histories for both downbursts and ABL, illustrating the instantaneous and mean wind speeds. In contrast to the ABL time history, which maintains a constant mean wind speed, the downburst time history is non-stationary, characterized by rapid increase, ramp-up, plateau, and decrease, ramp-down, in the wind speed. Therefore, for the downburst case a moving-mean approach is used to describe the time-varying mean. Figure 17 presents a comparison between the WOW downburst profile and various numerical and experimental data, demonstrating a strong match. Likewise, the ABL profile is compared to the ESDU open terrain profile, also showing a good match.

The pressure coefficient time history is calculated using Equation 1 by normalizing the measured pressure by the maximum mean wind speed which is the maximum moving mean in the case of downburst. Having said that, the mean wind speed for the ABL testing is 17 m/s while the maximum moving mean for the downburst testing is 10 m/s. C p t = P t − P 0 1 2 ρ U max ¯ 2 (1) where P t is the measured pressure, P 0 is the reference static pressure, U max ¯ is the maximum mean wind speed. Then, the mean pressure coefficient is the mean or maximum moving mean of the calculated time history in ABL and downburst cases, respectively.

Figure 18 shows a comparison between the mean pressure coefficients resulting from both downburst and ABL simulations for 0° deg wind direction. As depicted, the windward pressure coefficient follows the wind speed profile with stagnation area, corresponding to the highest positive pressure, occurs at higher elevation for ABL case while occurring at lower elevations for the downburst case matching the maximum wind speed of the nose-shape profile of the downburst. Despite the location, the maximum values of the pressure coefficient of the windward wall are comparable for both cases.

On the sides, the minimum negative pressure coefficient in the downburst case reaches 1.6 at the base of the leading edge, compared to 1.2 at the middle top in the ABL case which may have contributed to the increased observed damage in the case of the derecho compared to the hurricane especially for the total energies building, given the close range of expected wind gusts from the events as indicated in Sections 2.1, 2.2, which sustained considerable damage on its side wall (Kalliontzis et al., 2024). Additionally, the side pressures exhibit different characteristics. The downburst case has a decreasing magnitude gradient from the base of the leading edge towards the top of the trailing edge, matching the observations of Li et al. (2023), which is different from the typical pressure coefficient contour in the ABL case which shows the maximum negative values near the leading edge at the upper part and base of the side wall (Holmes, 2015). For the downburst case the highest suction in the leeward and side wall occurs at the same height of the max pressure on the windward. The leeward wall, in the ABL case is showing a radial contour plot with the maximum value located at the middle of the top region. In the downburst case, however, the leeward wall is showing different characteristics where the maximum suction is located at the edges at the same height of the stagnation area.

In addition to reporting the mean pressure coefficient, the characteristics of the fluctuating component should be studied to evaluate the turbulent nature of downbursts which affects the potential of inducing damage to tall buildings facades (Zhang et al., 2014). Figure 19 shows a comparison between the Root Mean Square (RMS) pressure coefficients resulting from both downburst and ABL simulations for 0° degree wind direction. In the downburst case, the RMS was calculated for the peak zone of the event which was approximately 3 s. For the windward wall, the RMS pressure coefficients range from approximately 0.09–0.22 in the downburst case, depicting a broader range compared to that observed in the ABL case, which ranges from 0.14 to 0.19. The higher values in the downburst case, especially in the lower half of the building, reflects the variation in the incident turbulence intensity between the two flow types where the turbulence intensity in the downburst case ranges from 0.12 to 0.17 along the height of the model which is higher than the values in the ABL which range from 0.08 to 0.14. On the side walls, the downburst case exhibits generally lower RMS values, except near the lower base close to the leading edge, where the downburst is showing higher values. The lower RMS values in the downburst case suggest that crosswind forces are weaker compared to the ABL case. For the leeward wall, the RMS pressure coefficients are higher at the sides in the ABL case, with values decreasing toward the center. A different pattern is observed in the downburst case where the highest RMS coefficient is located at the base of the leeward wall and decrease upward. Similar to the side walls, the values in the ABL case are higher than the those for downburst case. Hence, in general, the downburst showed lower RMS values except for the lower half of the windward wall and the base of the side walls. Overall, the values of the RMS pressure coefficient of the downburst case align closely with the results reported by Zhang et al. (2014). Additionally, the observed peak pressure coefficient is shown is Figure 20. The results are showing similar observation as the downburst demonstrates lower observed peaks except for the lower half of the windward wall and the base of the side walls.

For the 45° wind direction, illustrated in Figure 21, positive mean pressure coefficients of 0.9 are observed at the leading edge on the two wind-facing walls, face 1 and face 2, in both the ABL and downburst cases, with pressure decreasing towards the trailing edge. However, the downburst exhibits a steeper gradient of pressure decrease, resulting in higher suction at the trailing edge compared to the ABL case. Similar to the 0° wind direction, on the other two walls (face 4 and face 5), the downburst produces greater suction, reaching a mean negative pressure coefficient of −1.1, compared to −0.7 in the ABL scenario.

As observed, channelling effects in dense urban environment might have a significant consequence on the wind-induced local pressures and have contributed to the damage observed in Houston during the derecho. This section compares the mean pressure coefficients on the windward wall of the main model with different distances from an upstream interfering building of the same dimensions for both ABL and downburst winds. For ABL wind, Figure 22A shows the ABL-induced mean pressure coefficients on the windward wall for the isolated building while Figures 22B–F shows the mean pressure coefficient with upstream building while increasing its distance. Each case corresponds to increasing the upstream distance by the width of the model as depicted in Figure 15. Figure 22B is when the spacing between both buildings was equal to the width of the building. As shown, at small upstream distance, the windward wall is showing high negative pressure coefficient that reached a value of 0.7 with a bulk area of the windward subjected to this high suction. However, increasing the distance to twice the building width as in Figure 22C reduced the absolute coefficient to a value of 0.2 in most of the surface area despite having a value that reached 0.6 in very limited zone. This demonstrates that upstream distance dramatically influences pressure coefficients, with the original pattern gradually restored as distance increases since the flow gradually returns to the original state and the pressure coefficient stabilize and returns to the undisturbed case. Similar results were reported by Xie and Gu (2004) where the interference factor, defined as the ratio between the base moment of building with interference effects and the base moment of the isolated building, approaches 1 as the upstream distance increases. Figure 22F shows a very similar pattern, despite lower magnitudes, compared to the base case without interference effects. A similar trend is observed for the downburst case as shown in Figure 23. However, the original pattern and values in the downburst case is taking longer distance to be retrieved than the ABL case. The high suction in the windward wall with close upstream building might have contributed, beside wind-borne debris, to the façade damage of several buildings during the derecho such as the El Paso Energy Building and the CenterPoint Energy Plaza. A separate study should delve more to identify the difference in interference effects between ABL and downburst considering additional interference configurations and studying the effects on all surfaces for both mean and fluctuating components of the pressure coefficient.