Downburst related damages in Brazilian buildings: Are they avoidable?

Downburst related damages in Brazilian buildings: Are they avoidable?

Journal of Wind Engineering & Industrial Aerodynamics 185 (2019) 33–40 Contents lists available at ScienceDirect Journal of Wind Engineering & Indus...

4MB Sizes 2 Downloads 64 Views

Journal of Wind Engineering & Industrial Aerodynamics 185 (2019) 33–40

Contents lists available at ScienceDirect

Journal of Wind Engineering & Industrial Aerodynamics journal homepage: www.elsevier.com/locate/jweia

Downburst related damages in Brazilian buildings: Are they avoidable? Acir M. Loredo-Souza a, *, Elias G. Lima a, Matthew B. Vallis a, Marcelo M. Rocha a, Adri an R. Wittwer b, Mario G.K. Oliveira c a

Universidade Federal do Rio Grande do Sul, Av. Osvaldo Aranha, 99 / 305, Porto Alegre, RS, Brazil Universidad Nacional del Nordeste, Av. de las Heras, Resistencia, 3500, Argentina c Vento-S Consulting, Av. Di ario de Notícias, 400 / 1506, Porto Alegre, RS, Brazil b

A R T I C L E I N F O

A B S T R A C T

Keywords: Downbursts Thunderstorms Brazil wind code Wind-related disasters

Brazil's territory has shown an enormous potential to generate severe weather and intensive downbursts cases gained attention as a result of the severe destructive patterns not frequently seen before. However, several of the downburst damages on Brazilian buildings and structures follow a similar pattern than those caused by synoptic winds. The significant amount of accidents due to extreme winds in the recent years suggests the need for a review of the Brazilian Wind Code as well as the necessity to reinforce its use in common day buildings design. The new version of the code should incorporate an update of climate data and implement a thunderstorm wind model. This paper brings an overview of the downburst areas with major risk of occurrences, as well as presents an analysis of the damage patterns caused by the downbursts, with special focus on the effects on buildings and structures. It is concluded that, although the flow characteristics of downburst winds and conventional boundary layer simulations of synoptic winds may differ, the use of current conventional wind tunnel simulations and conventional code values are still a valuable tool in the prevention of wind related damages, at least in low and medium-rise buildings.

1. Introduction The frequency of weather-related disasters has increased when compared with any other kind of natural disasters on last decades (Guha-Sapir et al., 2011). In 2012, 93% of observed natural disaster were weather related and 45% of these events were caused by short-lived and small to meso scale atmospheric processes, including tornados or downbursts (Leaning and Guha-Sapir, 2013). These events caused innumerous transportation disruptions, power outages, infectious disease outbreaks and in the most extreme instances, loss of life (Leaning and Guha-Sapir, 2013; McMichael, 2015). In Brazil, 80% of natural disasters are related to severe atmospheric instabilities and 94% of these events have been reported to have occurred in the most populous areas of the country: South, Southeast and Northeast (Loredo-Souza, 2012; Marcelino et al., 2005). For the South and Southeast regions this is due to climatic characteristics, but for the Northeast this is because of the presence of people reporting the events. Population growth, socio-spatial segregation, capital accumulation and protected areas invasions remain as the main reasons that lead

populations to get more and more sensitive to extreme weather conditions (Marcelino, 2007). Consequently, preventive and energetic government actions must be taken in order to improve population safety and private/public property protection. Brazil has a great part of its territory located on mostly tropical areas, with high convection and humidity combined with intense cold fronts coming from the south; these serve as initiators of severe weather in the country. Consequently, the hot season on the Southern Hemisphere is found to be the most propitious time to severe weather development (Romatschke and Houze, 2010). The electricity on the atmosphere also indicates that the austral spring and the summer seem to be much more active than the other seasons, reaching a maximum in January and a minimum in July (Anselmo, 2015). The north part seems to have the biggest annual lighting density. The center-west, southeast and south parts of Brazil have also high values of this index. Rio de Plata Basin, which cover the western half part of Brazil and its South Area, is highlighted by having the highest atmosphere electricity efficiency, producing from 195% to 323% more thunderstorms than the rest of the area with the same annual lighting density (Anselmo, 2015). The high

* Corresponding author. Rua Ferreira Viana, 14/5, Porto Alegre, RS, CEP: 90.670-100, Brazil. E-mail addresses: [email protected] (A.M. Loredo-Souza), [email protected] (E.G. Lima), [email protected] (M.B. Vallis), [email protected] (M.M. Rocha), [email protected] (A.R. Wittwer), [email protected] (M.G.K. Oliveira). https://doi.org/10.1016/j.jweia.2018.11.022 Received 30 June 2018; Received in revised form 15 October 2018; Accepted 15 November 2018 0167-6105/© 2018 Elsevier Ltd. All rights reserved.

A.M. Loredo-Souza et al.

Journal of Wind Engineering & Industrial Aerodynamics 185 (2019) 33–40

frequency of thunderstorms on the South Area of Brazil is due to the formation of a low-level circulation east of the Andes. This deep convection system is promoted on subtropical latitudes of South America (Nascimento, 2005), where Low Level Jets (LLJ) contribute to the atmospheric destabilization through moisture transportation from the Amazon Region, especially during the austral spring and autumn. With the occasional coupling between LLJ and High Level Jet (HLJ), an important dynamic mechanism is established and deep and low pressure systems are commonly developed generating severe storms events. A similar system is observed in North America and in the same way it can generate small spatial and time scale weather phenomena, such as downbursts (see Fig. 1). Downbursts are associated with thunderstorms and are highlighted among the most severe phenomena to be observed in nature (Fujita, 1985). It was first identified by Foster (1958), but at that point it was still mistaken as gust front events. It was only in 1985 that Fujita controversially first coined the term Downbursts, describing it as a strong dense column of cold air, a downdraft, which descends towards the ground, inducing strong burst and divergent winds, also known as an “outburst” (Fujita, 1985). Today the concept of downburst is very well established, and it is ideally defined as shown at Fig. 2 (Hjelmfelt, 1988). In this case, it is assumed the flow structure is completely developed to its maximum, presenting zones in which might be found maximum wind speed values. In terms of vertical variation, the velocity profile of a downburst event (in the ground spreading process) differs from the typically observed atmospheric boundary layer (ABL) profiles as indicated in Fig. 3. This may be paramount for very tall buildings, but for those studied in this work (not taller than 80 m) perhaps it is not a big issue. However, little is known regarding the turbulence structure of the downburst flow and differences may appear. As seen, thunderstorm winds (TS) present a vertical mean velocity profile with a different shape than that traditionally encountered on Extended Pressured System (Ponte and Riera, 2010), although in its bottom part it resembles a boundary layer velocity profile. There are many doubts, however, regarding its structure of turbulence. Some authors tend to point out that thunderstorm events present a smaller occurrence probability, but the damage magnitude caused by TS events on the lasts decades have been a constant concern of designers and engineers, and has become a focus of several investigations, especially after the 1980's (Riera and Nanni, 1989). Fields tests have been performed at the Wind Science and Engineering Research Center of Texas Tech University to evaluate full-scale nonstationary correlated downburst wind speeds and then, to develop a methodology to model downburst vertical profiles (Chen and Letchford, 2005, 2006, 2007). Wind speed time series of downbursts were obtained on June 4 and 15, 2002. Two sets were recorded simultaneously at different heights and at seven observation towers. The statistical procedures utilized in the work of Chen and Letchford (2005) allowed

Fig. 2. Transverse section of a typical downburst structure (Hjelmfelt, 1988).

observing that the time-varying standard deviations of fluctuating wind speeds may be approximately proportional to the time-varying mean speeds (Chen and Letchford, 2006). The importance of understanding the downburst incidence in Brazil (Garstang et al., 1998) becomes evident when analyzing the losses and damages associated with them (Lima and Loredo-Souza, 2015; Loredo-Souza et al., 2016). The significant amount of accidents due to extreme winds in the recent years suggests the need for a review of the Brazilian Wind Code (NBR-6123, 1988)(which is presently under way) as well as the necessity to reinforce its use in common day buildings design. It is aimed that this new version of the code incorporates an update of climate data (Vallis et al., 2017) and implements a thunderstorm wind model. This paper brings an overview of the downburst areas with major risk of occurrences, including the South and Southern Region, Western Amazon Basin, and the Northeast Coast. It also presents an analysis of the damage patterns caused by the downbursts, with special focus on the effects on buildings and structures. 2. Identifying downbursts Convective indexes can indicate if convective systems are favorable to be developed, raising the possibility of downburst occurrence. Convective Available Potential Energy - CAPE is the positive buoyancy of an air parcel and is an indicator of atmospheric instability. CAPE values from 1000 to 2500 m2/s2 are considered high, values above 2500 m2/s2 indicate a pronounced instability and values above 4000 m2/s2 indicate extreme instability. In addition, Convective Inhibition – CIN indicates the amount of energy required to overcome the negatively buoyant energy the environment exerts on an air parcel. CIN values between 50 and 100 m2/s2 are considered to indicate enough convective inhibition, but when combined with high CAPE values indicate extreme convective

Fig. 1. Common dynamic structure interaction between Low Level Jets and Subtropical Jet (SJ) and Polar Jet (JP) High Level Jets during periods of increased convective activity in (a) South America and (b) North America (Nascimento, 2005).

Fig. 3. Schematic of an atmospheric boundary layer profile, on the left, and the velocity profile in the outflow of a downburst, on the right (Bertsch and Ruck, 2015). 34

A.M. Loredo-Souza et al.

Journal of Wind Engineering & Industrial Aerodynamics 185 (2019) 33–40

environment (Caracena et al., 1989). Atmospheric vertical soundings also contribute to identify downburst development environments. A dry layer in mid-tropospheric levels, between 700 hPa and 500 hPa can suggest potential energy to originate a strong downward (Gilmore and Wicker, 1998). Due to the entrainment of cold and dry air at high levels, the formation of low Equivalent Potential Temperature θe and extreme dry air can ensure a negative buoyancy and downdraft trajectory out of the cloud, causing a mesohigh area at ground level. In addition, a significant vertical wind shear ensures a feedback to the storm to keep the system going on (Atkins and Wakimoto, 1991). Observations of microbursts near to Chicago and in South Florida indicated that in cases which θe differences between the surface and the minimum value in mid-tropospheric levels was greater than or equal to 20 K, presented a greater potential to the downburst occurrence. For lower values of 13 K, intense storms could be formed, but without downbursts (Atkins and Wakimoto, 1991). Gusts observations greater than 10 m/s are initial conditions to consider the occurrence of downbursts, according to Garstang et al. (1998). The maximum observed in their study at Amazon Forest was 17.2 m/s, however, these velocities values can still not explain the extended damages of trees, suggesting that necessaries forces to pull-out those elements would require a maximum speed around 25 m/s, which probably occurred, but were just not registered (Garstang et al., 1998). Some authors indicate a 30% increase to the registered wind gusts (Fujita and Wakimoto, 1981). It was observed that gusts produced by downbursts are usually accompanied by precipitation, an abrupt decrease of the equivalent potential temperature (θe) at the measurement level, ranging between 4.00 K and 18.74 K besides an instantaneous decay of specific humidity greater than 3.5 g/kg air (Garstang et al., 1998). Fujita, in his first experiment, observed a pressure increase of 4hPa on a 2 km microburst nose [15]. Other studies identified 5 hPa increase in the vortex nose region (Garstang et al., 1998) and 2.4 hPa on the Florida Area Cumulus Experiment – FACE (Caracena and Maier, 1987). The most frequent time found for downbursts observations is between 8 a.m. and 8 p.m. (local), with a mean time around 1 p.m. (local), this is the part of the day in which occurs the most local convective activity (Garstang et al., 1998). To better understand the meteorological parameters performance on a downburst occurrence, Fig. 4 shows an event registered at Amazonas Rainfall. Table 1 presents a range of meteorological parameters available in the literature that indicate several values during downbursts occurrence on convective environments.

Table 1 Range of values of meteorological parameters that indicate the occurrence of downbursts on convective environments taken in account on this study. Characteristics

Range of values

Reference

Decrease of Equivalent Potential Temperature (θe) between surface and the coldest layer near 700hPa (K) Wind Gusts (m/s)

>20

Leaning and Guha-Sapir (2013) Garstang et al. (1998) Foster (1958) Garstang et al. (1998) Garstang et al. (1998) Loredo-Souza et al. (2016) Garstang et al. (1998) Garstang et al. (1998) Garstang et al. (1998) Foster (1958)

10 (Minimum) 25 (Mean) >4

Decrease of Equivalent Potential Temperature (K) Temperature Decrease ( C)

>5

Increased Atmospheric Pressure (hPa)

>2,4

Drop in Mixing Ratio (g/kg)

>3,5

Start of the precipitation along the downdrafts (mm/day) Meteorological phenomenon that originated the downburst Reflectivity radar (dBZ)

>0,5 Supercells, squall lines and Derechos Dry >15 Wet <65

3. Case studies Next, the parameters for the downburst occurrence are analyzed for two specific cases that occurred in 2012 and 2013, and then the damages caused by some downbursts that occurred in Brazil since 1990 are described.

3.1. Analysis of parameters for the downburst occurrence In order to test these parameters for the downburst occurrence, two specific analyses were made, the first one occurred on the night of December 31st, 2012, and the second one, during the day on May 29th, 2013. Both cases were chosen due to weather severity observed typical for downbursts events. These cases were triggered by the presence of Mesoscale Systems and both were intensively noticed because of the strong social impact on Rio Grande do Sul (RS) and Santa Catarina (SC) States. For this study, it was used hourly data from automatic weather stations of Nacional Institute of Meteorology - INMET and atmosphere vertical sounding taken in the cities of Porto Alegre and Santa Maria. On the first case, the atmosphere was intensely unstable all over RS State, clouds presented an intensive vertical development, with a deep dry air layer at mid tropospheric levels enhancing the evaporative cooling which is responsible for the development of cool and dense air of downbursts occurrence. The observed data on that event is available on Table 2 and when related with the available data found on Table 1, 81% of the parameters (underlined with a double line) are equal or exceed the suggested by the literature found with downbursts occurrences. The values of CAPE and CIN index recorded on Porto Alegre at 00Z of January 1st, 2013, were 1132 J/kg and 137 J/kg, respectively. In Santa Maria it was found 1648 J/kg and 38.1 J/kg, respectively. The indexes point out intensive instability conditions, suggesting the possibility to Downbursts have occurred. The second case studied was on May 29th, 2013, it was associated with strong damages with the progress of a squall line over the RS state, which advanced through the southern part of the SC State, causing major damages to the local population. The data obtained in the second case is presented in Table 3, 50% of the values (underlined with a double line) are higher than the suggested by literature to characterize the downburst occurrence. This case was shown less expressive than the first one. The θe variation for the two cases at the soundings moments on May 29th, 2013, at 12 Z and 30th, 2013, at 00 Z were shown below and they highly suggest

Fig. 4. Registers made at 5 m height above the treetops, on April 23, 1987 on a Weather Station at Amazon Rainforest. (a) Wind speed, (b) Equivalent Temperature, (c) Precipitation, (d) Specific Humidity and (e) Air Temperature (Garstang et al., 1998). 35

A.M. Loredo-Souza et al.

Journal of Wind Engineering & Industrial Aerodynamics 185 (2019) 33–40

Table 2 Data occurred on the evening of December 31st, 2012, in southern Brazil. Local

Date

Hour (UTC)

Vmax m/s

Vmax/Vmin (%)

T max - T min (k)

P max - P min (hPa)

r max - r min (g/kg)

θe max - θe (K)

Rio Grande Santa Maria Santana do Livramento Uruguaiana Canguçú Caçapava do Sul Alegrete Bage Quaraí S~ ao Gabriel Juguar~ ao Don Pedrito Chuí

1/1/13 1/1/13 12/31/12

0000 0200 2300

18,70 17,90 21,80

47 8 33

6,40 10,50 9,20

4,30 5,40 4,60

3,6 17,6 14,6

17,35 23,45 21,45

24 33 30

28,6 34,4 20,2

12/31/12 1/1/13 1/1/13 1/1/13 12/31/12 12/31/12 1/1/13 12/31/12 12/31/12 12/31/12

2300 0100 0100 0100 2300 2300 0000 2200 2300 2100

23,00 22,40 18,80 19,40 19,20 28,80 18,00 16,00 19,00 23,10

29 32 45 29 22 24 28 45 70 13

10,90 0,40 0,70 2,60 8,80 9,60 8,20 3,50 8,00 5,20

4,40 2,40 3,60 2,80 3,50 2,90 2,20 2,80 4,70 4,60

19,5 0,6 1,1 4,2 14,2 16,9 5,7 5,6 13,1 7,5

28,92 0,87 0,74 5,62 21,84 19,22 26,30 9,34 24,84 17,02

38 2 5 13 28 36 15 14 25 14

17,6 23,2 13,0 17,0 1,4 19,8 7,4* 11,4 22,6* 21,4

20,5 3,2

33 0,2

6,5 3,5

3,7 1,0

9,6 6,4

16,7 9,2

21,3 11,2

18,90 8,3

min

UR max - UR (%)

min

Prec. (mm)

*preciptating on the next hour Mean Standard Deviantion

Table 3 Data from the second study of multiple cases for the case occurred on May 29th, 2013. Local

Hour (UTC)

V

Canela Canguçú Uruguaiana Rio Grande Rio Pardo Santa Maria Tramandaí S~ ao Gabriel Torres Vacaria S~ ao Jose dos Ausentes Porto Alegre

17:00 22:00 11:00 16:00 15:00 14:00 17:00 14:00 18:00 19:00 19:00 15:00

Ararangu a S~ ao Jose Urussanga Morro da Igreja Santa Marta S~ ao Joaquin Ituporanga

19:00 23.00 21:00 21:00 20:00 21:00 23:00

Mean Standard Deviantion

m/s

r max - r min (g/kg)

θe max - θe (K)

3,5 2,9 8,4 1,6 4,1 4,3 4,8 8,9 2,7 0,9 1,5 4,3

3,40 2,48 1,71 0,39 0,56 3,27 1,59 1,81 1,53 0,51 3,62 0,47

15,05 11,73 7,53 1,59 1,86 13,95 6,33 6,28 6,08 4,23 17,19 1,57

11 2 8 1 7 11 3 8 4 19 13 1

9,0 1,0 10,6 8,2 5,8 4,0 12,6 5,8 7,6 15,4 11,0 4,2

3,4 3,6 4,1 4,3 4 0,7 4,5

4,7 2,7 1,7 1,7 1,9 1 1,9

1,23 1,50 2,04 2,93 0,80 0,41 1,73

6,83 8,01 10,21 13,34 6,59 1,86 9,77

14 17 28 5 10 7 30

20,6 6,4 0,6 1,2 2,2 3,8 8,6

3,0 1,6

3,3 2,2

1,7 1,0

7,9 4,6

10,5 8,1

7,3 5,1

Vmax/Vmin (%)

T max - T (k)

16,9 20,7 21,5 15,1 22,5 25,4 23,6 18,5 15,3 17,4 22,9 9,8

27 37 35 56 24 40 20 43 16 22 47 26

4,9 4,7 3,2 0,6 0,6 4,5 2,1 1,9 1,8 2,5 5,9 0,6

17,8 11,6 14,5 30,6 31,7 18,5 10,2

14 31 8 64 20 26 4

19,2 5,9

0,3 0,2

max

P max - P (hPa)

min

min

min

UR max - UR (%)

min

Prec. (mm)

A very impressive downburst event occurred in January 29, 2016, right at the urban perimeter area of Porto Alegre, one of the major cities in Brazil. The winds caused damage in the majority of the city, leaving more than 220,000 houses without electricity and thousands without water. Brazilian meteorological services indicated that an approaching cold front encountering moist warm air led to the formation of a supercell over the whole metropolitan area. The temperature reached 40  C, which is normal for summer in Porto Alegre and usually leads to the formation of thunderstorms and extreme winds (Loredo-Souza, 2012), but this particular event had a longer duration (more than 30 min) and sustained high wind speeds. Meteorologists and Wind Engineers had a common understanding that this severe event was a downburst, and more specifically, a macroburst. The maximum gust measured was 33.2 m/s, at the INMET Station. At the airport, the maximum gust recorded was 24.2 m/s and at the downtown harbor the maximum measured gust was 27.2 m/s. From the damage characteristics, meteorologists estimate that over large areas the wind speed reached around 28 m/s and in a few neighbourhoods 42 m/s. Trees and cars were knocked-over throughout the city. A non-whirl pattern was observed and nothing was really thrown up, dropping off

downburst occurrence, however the data recorded at meteorological station of Morro da Igreja, in S~ao Joaquim, SC, and at in Santa Marta Lighthouse, in Laguna, SC, had shown values around to what is available on literature to define the downburst occurrence. The registered wind gusts were 28 m/s on those cases, and they might be result of a downburst generated wind gust, just as the cases in S~ao Jose dos Ausentes, RS, and Santa Maria, RS, which were also extremely high. CAPE and CIN indexes recorded in Porto Alegre at 12Z on May 31st, 2013, and 1.61 J/ kg and 408 J/kg, respectively, and Santa Maria was 275.1 J/kg and 269 J/kg, respectively. These values where not significant to define a downburst environment, highlighting the limitations of convective index for thunderstorms prediction. 3.2. Description of observed cases of downburst Reports regarding downbursts occurrence in Brazil were found since the 1990's and some of them are listed in Fig. 5 according to their geographical distribution in Brazil along with the indicated cases analyzed in this study. Further case studies will be presented in following publications. 36

A.M. Loredo-Souza et al.

Journal of Wind Engineering & Industrial Aerodynamics 185 (2019) 33–40

developed above the city at the same time the strongest wind gusts were registered. As Garstang et al. (1998) suggested, supercells are typical weather systems that might be able to generate downbursts. But there are several meteorological parameters, such as those shown in Table 1, that can be used to describe a typical range found during a downburst occurrence (Lima and Loredo-Souza, 2015). Also presented are the corresponding characteristic values obtained during the event in analysis, where is possible to verify the abrupt variability of the predominant meteorological parameters (Table 4). Due to data limitations of the Automatic Weather Stations of Brazilian Meteorological Service, for Equivalent Potential Temperature and Mixing Ratio the instant values of the hour interval in which the downburst was registered and the previously registered hour were used. For other parameters, maximum and minimum amplitudes in the hour of the event were used, and they all describe a sudden change on ambient during event occurrence.

Fig. 5. Downbursts observed cases in Brazil in an initial study. Red areas represent cases reported in the literature and purple areas suggested cases indicated by this study.

the hypothesis of a tornado. Besides, videos and eyewitnesses report strong downward winds followed by horizontal winds. Several damages occurred in buildings, particularly to the façades. Fig. 6 shows some examples of failures occurred during the event. This meteorological event offered an opportunity to check and compare, in full-scale and under a downburst flow, the performance of buildings previously tested in a conventional Boundary Layer Wind Tunnel. Fig. 7 shows some pictures of the collapse of a 37 story building in the city of Belem, state of Para, in the Amazon Region in Brazil, due to a downburst event occurred on January 29, 2011. The building construction and finishing was almost ready, the accident happened on a Saturday and the number of deceased people could have been much higher. Fig. 8 brings a set of pictures of full-scale damage caused by a downburst occurred in the city of Campinas, S~ao Paulo state, on June 5, 2016.

Fig. 7. Collapse of a 37 stories building in the city of Belem, state of Para, in the Amazon Region in Brazil, due to a downburst event occurred on January 29, 2011. The left top figure shows the building before the downburst and the left bottom figure was taken after the collapse. The figures in the right show the instant of the downwash (top), the after crash (middle) and people being rescued (bottom).

4. The Porto Alegre event 4.1. Characteristic parameters of the downburst event A downburst event is a very unique meteorological phenomenon and several factors need to be observed before confirming its occurrence. The severe weather event observed in Porto Alegre can be defined as a macroburst mainly due to the pattern of destruction, but if radar images were available with a better and reliable resolution the event could be better understood. The satellite image of Fig. 9 shows a supercell well

Fig. 8. Full-scale damage caused by a downburst occurred in the city of Campinas, S~ao Paulo state, on June 5, 2016. The figures mainly show the external and internal damages occurred in roofs.

Fig. 6. Examples of damages occurred during the Porto Alegre January 29, 2016, downburst event. From top left, clockwise: (a) gas-station roof structure and cladding failure, (b, c) buildings glass cladding failure, (d) building internal damage, (e) boat partially sinked.

Fig. 9. Satellite image of the supercell over Porto Alegre on 30/01/2016 00:45:51 UTC (REDEMET, 2016). 37

A.M. Loredo-Souza et al.

Journal of Wind Engineering & Industrial Aerodynamics 185 (2019) 33–40

Table 4 Typical meteorological parameters values for the downburst observed for Porto Alegre Weather Station (A801) during the event in analysis. Characteristics

Porto Alegre

Decrease between surface and colder layer near the 700hPa (K) Wind gust (m/s) Effective decrease of instant equivalent potential temperature (K) Temperature decrease ( C) Dew point decrease ( C) Atmospheric pressure increase (hPa) Saturated instant mixing ratio decreasing (g/kg) Relative humidity decrease (%) Registered precipitation along downdrafts (mm/h)

Sounding data not available 33.2 14.93 5.5 1.8 2.4 1.70 22 37.4

As shown in Table 5, some weather stations around Porto Alegre also presented wind gusts higher than 10 m/s, but only Campo Bom (A884), 42 km from Porto Alegre, presented substantial pressure peak, followed by strong temperature decrease, bringing to light another possible downburst occurrence on the “Great Porto Alegre” area. Differently, Canela (A879), 85 km distant, and Bento Gonçalves (A840), 100 km distant, had strong wind gusts, but no pressure peak. Temperature reductions might be justified with the approaching of cold air with a front gust, explaining also the wind direction change. The data available on Table 4 is really relevant for this study because all but one typical meteorological parameters for a downburst occurrence were registered on the event of January 29th, 2016, confirming the hypotheses that such strong phenomena stroked the urban area of Porto Alegre. Previous studies have already indicated that the southern part of Brazil is susceptible to downburst occurrence (Lima and Loredo-Souza, 2015). The main concern about the recent events remains in regard to building safety, since it is not yet fully understood in which extent the wind characteristics generated by downbursts are different to those of the Extended Pressure Systems, or synoptic winds (Letchford and Chay, 2002). This may leave structures susceptible to failures, risking lives and causing major economic losses, as those which were observed in the last downburst event in Porto Alegre.

Fig. 10. Models inside the wind tunnel: two soccer stadiums and six 80 m high buildings.

The wind tunnel tests were carried out including the physical model of the incident wind corresponding to the type of the local terrain of the tested structure. A description of the characteristics of boundary layer winds that can be simulated in the UFRGS tunnel is indicated in Blessmann's previous work (Blessmann, 1982). Additionally, the surrounding topography and the near urban environment were reproduced on the same model scale of the tested buildings. The models were constructed with a scale of 1: 400 using the 3D print technique and were tested with incident wind velocity of approximately 35 m/s. All models were instrumented with pressure taps and instantaneous pressure measurements were taken over the whole models surfaces. Fig. 11 shows the mean pressure coefficient distribution for one of the buildings, and direct wind incidence. Due to space limitations, the pressure coefficient distributions for other buildings are shown in the next section, together with the damage configurations. 4.3. Full-scale performance A map with more than 400 cases of damages through the city, registered by Metroclima (2016), is shown in Fig. 12 together with the places where official velocity measurements were available. The damages are separated, according to the icons: felled trees, structural/cladding damage (gears) and fire. The density was higher than the map is presenting, but for clarity only the stronger damages are shown. Also indicated in Fig. 12 are the locations of the structures for which the wind tunnel tests results were available.

4.2. Wind tunnel tests This meteorological event offered an opportunity to check and compare, in full-scale and under a downburst flow, the behaviour of buildings previously tested in a conventional Boundary Layer Wind Tunnel. Six 80 m high buildings and two soccer stadiums were analysed. Fig. 10 shows the models inside the Prof. Joaquim Blessmann Boundary Layer Wind Tunnel of the Universidade Federal do Rio Grande do Sul. Table 5 Meteorological parameter obtained with the weather stations around Porto Alegre. Characteristics

Canela

Campo Bom

Bento Gonçalves

Decrease between surface and colder layer near the 700hPa (K) Wind gust (m/s) Effective decrease equivalent potential temperature (K) Temperature decrease ( C) Dew point decrease ( C) Atmospheric pressure increase (hPa) Saturated instant mixing ratio decreasing (g/kg) Relative humidity decrease Registered precipitation along downdrafts (mm/h)

Sounding Data Not Available 17.3 17.41

Sounding Data Not Available 13.8 18.84

3.7 4.4 1.2

7.6 5.0 3.7

Sounding Data Not Available 12.2 No decrease registered 1.4 0.7 1.0

4.35

3.67

8 0.6

29 6.4

No decrease registered 11 0

Fig. 11. Mean pressure coefficient diagrams on buildings façades from boundary layer wind tunnel tests (suction ¼ red, yellow; pressure ¼ blue) for the central building indicated. 38

A.M. Loredo-Souza et al.

Journal of Wind Engineering & Industrial Aerodynamics 185 (2019) 33–40

Fig. 14. Full-scale damage to surroundings caused by downburst and no damage to stadium.

of the wind tunnel results into structural design and construction considerations. For the group of buildings, the performance was different. All of them presented damage in their façades (Figs. 15–17), but in very distinct patterns. For the first group of three buildings shown in Fig. 11, for which some pressure coefficients diagrams are also presented, the evidences show that for the windows that were kept closed and locked did not suffer any damage, while some of those which were open, or even closed but not locked, were ripped out or broken by the wind action (Fig. 15). In three of the most affected buildings (Figs. 16 and 17) the façades had two types of cladding elements: glass and granite plates. It is possible to observe that the most severe damages occurred in the granite cladding, which has a specific support configuration (Loredo-Souza et al., 2016) as indicated in Fig. 18. Very interesting to note is that the pressure distribution diagrams on the building façades, obtained from the wind tunnel study, show a direct correspondence with the damaged zones of the analyzed full-scale buildings. For Fig. 17 the correspondence between pressure distribution and damage to the granite cladding is very evident, even that there is no

Fig. 12. Map indicating the damages suffered in the Porto Alegre January 29, 2016, downburst event, the places and values of the official velocity records and the locations and buildings for which boundary layer wind tunnel results were available (adapted from Metroclima, 2016).

4.4. Full-scale performance Regarding the wind velocities, even if the worst estimative is confirmed, resulting in larger velocities than those effectively measured, the reference wind gust velocity recommended for Porto Alegre by the Brazilian Wind Code is 46 m/s (Fig. 13) (NBR-6123, 1988). This means that the resulting damages were not caused by a non-predicted phenomenon, but maybe for a possible misunderstanding in the design indications and/or operational conditions of the structures and cladding. Analysing the performance of the structures it is possible to verify that the stadiums remained undamaged, while the trees and surrounding structures were seriously damaged as can be seen in Fig. 14. During the design stage, there were modifications and improvements in the stadium structures due to the wind tunnel tests results. Therefore, the lack of damages in the stadiums is believed to be due to the correct employment

Fig. 15. Full-scale damage to cladding elements: some windows that were not locked were destroyed.

Fig. 16. Full-scale damage caused by downburst and mean pressure coefficient diagrams on buildings façades from boundary layer wind tunnel tests (suction ¼ red, yellow; pressure ¼ blue) for Tower A.

Fig. 13. Map from the Brazilian Wind code with the 3s gust reference wind speeds, in m/s, at10 m height, open terrain, 50 years return period. The reference velocity for Porto Alegre is highlighted (NBR-6123 (1988)).

Fig. 17. Full-scale damage caused by downburst and mean pressure coefficient diagrams on buildings façades from boundary layer wind tunnel tests (suction ¼ red, yellow; pressure ¼ blue) for Tower C. 39

A.M. Loredo-Souza et al.

Journal of Wind Engineering & Industrial Aerodynamics 185 (2019) 33–40 Atkins, N.T., Wakimoto, R.M., 1991. Wet microburst activity over the southeastern United States: implications for forecasting. Weather Forecast. 6 (4), 470–482. https://doi. org/10.1175/1520-0434(1991)006<0470:WMAOTS>2.0.CO;2. Bertsch, A., Ruck, B., 2015. Interaction between convective downdrafts and inner city areas – a wind tunnel study. In: Proceedings of the 14th International Conference on Wind Engineering – ICWE14. Porto Alegre, Brazil. June 21-26. Blessmann, J., 1982. The boundary layer wind tunnel of UFRGS. J. Wind Eng. Ind. Aerod. 10, 231–248. Caracena, F., Holle, R.L., Doswell, F.A., 1989. Microbursts—a Handbook for Visual Identification. US Dept. of Commerce. NOAA/ERL/NSSL. Caracena, F., Maier, M.W., 1987. Analysis of a microburst in the FACE meteorological mesonetwork in southern Florida. Mon. Weather Rev. 115 (5), 969–985. Chen, L., Letchford, C.W., 2005. Proper orthogonal decomposition of two vertical profiles of full-scale nonstationary downburst wind speeds [lzcl]. J. Wind Eng. Ind. Aerod. 93, 187–216. Chen, L., Letchford, C.W., 2006. Multi-scale correlation analyses of two lateral profiles of full-scale downburst wind speeds. J. Wind Eng. Ind. Aerod. 94, 675–696. Chen, L., Letchford, C.W., 2007. Numerical simulation of extreme winds from thunderstorm downbursts. J. Wind Eng. Ind. Aerod. 95, 977–990. Foster, D.S., 1958. Thunderstorm gusts compared with computed downdraft speeds. Mon. Weather Rev. 86, 91–94. https://doi.org/10.1175/1520-0493(1958)086<0091: TGCWCD>2.0.CO;2. Fujita, T.T., 1985. The Downburst: Microburst and Macroburst. SMRP-rp-210. Univ. of Chicago, p. 122. Fujita, T.T., Wakimoto, R.M., 1981. Five scales of airflow associated with a series of downbursts on 16 July 1980. Mon. Weather Rev. 109 (7), 1438–1456. Garstang, M., White, S., Shugart, H.H., Halverson, J., 1998. Convective cloud downdrafts as the cause of large blowdowns in the Amazon rainforest. Meteorol. Atmos. Phys. 67, 199–212. https://doi.org/10.1007/BF01277510. Gilmore, M.S., Wicker, L.J., 1998. The influence of mid-tropospheric dryness on supercell morphology and evolution. Mon. Weather Rev. 126 (4). Guha-Sapir, D., Hoyois, P., Below, R., 2011. Annual Disaster Statistical Review 2010: The numbers and trends. Centre for Research on the Epidemiology of Disasters (CRED), Institute of Health and Society and Universite Catholique de Louvain: Louvain-laneuve, Belgium. Hjelmfelt, M.R., 1988. Structure and life cycle of microburst outflows observed in Colorado. J. Appl. Meteorol. 27, 900–927. https://doi.org/10.1175/15200450(1988)027<0900:SALCOM>2.0.CO;2. Leaning, J., Guha-Sapir, D., 2013. Natural disasters, armed conflict, and public health. N. Engl. J. Med. 369, 1836–1842. https://doi.org/10.1056/NEJMra1109877. Letchford, C.W., Chay, M.T., 2002. Pressure distributions on a cube in a simulated thunderstorm downburst. Part B: moving downburst observations. J. Wind Eng. Ind. Aerod. 90 (7), 733–753. Lima, E.G., Loredo-Souza, A.M., 2015. Analysis of downburst occurrence in Brazil. In: Proceedings of the 14th International Conference on Wind Engineering. Porto Alegre, Brazil. Loredo-Souza, A.M., 2012. Meteorological events causing extreme winds in Brazil. Wind Struct. An Int. J. 15, 177–188. https://doi.org/10.12989/was.2012.15.2.177. Loredo-Souza, A.M., Lima, E.G., Vallis, M.B., Rocha, M.M., Wittwer, A.R., Oliveira, M.G.K., 2016. Full-scale downburst damage versus boundary layer wind tunnel pressures: a survey analysis. In: Proceedings of the 8th International Colloquium on Bluff Body Aerodynamics and Applications. Northeastern University, Boston, Massachusetts, USA. June 7–11. Marcelino, E.V., 2007. Desasters Naturais e Geotecnologias: conceitos Basicos. Inst. Nac. Pesqui. Espac. 20. Marcelino, I.P.V. de O., Nascimento, E. de L., Ferreira, N.J., 2005. Tornadoes in Santa Catarina State (southern Brazil): event documentation, meteorological analysis and vulnerability assessment. Lagavulin.Ltid.Inpe.Br 1–15. McMichael, A.J., 2015. Extreme weather events and infectious disease outbreaks. Virulence 6, 543–547. https://doi.org/10.4161/21505594.2014.975022. Metroclima, 2016. https://pbs.twimg.com/media/CaYE610WYAAcBJ0.jpgINMET. Nascimento, E.L., 2005. Previs~ao de tempestades severas utilizando-Se par^ametros convectivos E modelos de mesoescala: Uma Estrategia operacional Adotavel No Brasil? Rev. Bras. Meteorol. 20, 121–140. Ponte, J., Riera, J.D., 2010. Simulation of extreme wind series caused by thunderstorms in temperate latitudes. Struct. Saf. 32, 231–237. https://doi.org/10.1016/j.strusafe. 2010.02.002. REDEMET, 2016. In: thttp://www.redemet.aer.mil.br/?i¼produtos&p¼imagens-desatelite. Riera, J.D., Nanni, L.F., 1989. Pilot study of extreme wind velocities in a mixed climate considering wind orientation. J. Wind Eng. Ind. Aerod. 32, 11–20. https://doi.org/ 10.1016/0167-6105(89)90012-3. Romatschke, U., Houze, R.A., 2010. Extreme summer convection in South America. J. Clim. 23, 3761–3791. https://doi.org/10.1175/2010JCLI3465.1. Vallis, M.B., Loredo-Souza, A.M., Watrin, L.C., 2017. A review of Brazilian wind data. In: Proceedings of the 13th Americas Conference on Wind Engineering (13ACWE) Gainesville, Florida USA, May 21-24.

Fig. 18. Full-scale damage caused by downburst and granite plates supporting system.

evidence that the design wind code velocity has been reached. The granite plates supporting system shown in Fig. 18 does not seem accordingly designed to withstand wind pressures from the magnitude reached in the event, remembering that the design wind code velocity was not reached. Regarding the glass cladding, some damages occurred, but in a much smaller amount, being some ripped out when left open or broken due to the impact with windborne debris. 5. Summary and conclusions This paper brings a review of downburst occurrence cases in Brazil, it points out how to identify environments in which occurrence might be observed in the country and the importance of considering this phenomenon in building design. It was found that in the eastern and western areas of Amazon Basin, in the Southern, and in the Southeast Regions of Brazil downbursts are frequently observed. Additionally, due to the influence of trade winds over the northeast, lines of instability are created, and might be the responsible for originating downbursts. However, it is denoted that this is a preliminary study and needs further investigation, bringing suggestions for future studies, a first step to better understand the downburst occurrence in Brazil and the importance of including in the wind code some indication of thunderstorm wind models. Brazil's territory has shown an enormous potential to generate severe weather and intensive downbursts cases gained attention as a result of the severe destructive patterns. However, several of the downburst damages on Brazilian buildings and structures follow a similar pattern than those caused by synoptic winds. Then, it may be concluded that, although the flow characteristics of downburst winds and conventional boundary layer simulations of synoptic winds may differ, the use of current conventional wind tunnel simulations and conventional code values are still a valuable tool in the prevention of wind related damages, at least in low and medium-rise buildings. Acknowledgements The authors acknowledge the support of the Brazilian National Research Council, CNPq. References Anselmo, E.M., 2015. reportMorfologia das tempestades Eletricas na America do Sul. Dissertation, Universidade de S~ao Paulo. S~ao Paulo, Brazil. Associaç~ ao Brasileira de Normas Tecnicas, 1988. NBR-6123 – Forças devidas ao vento em edificaç~ oes (Brazilian Wind Code). Rio de Janeiro.

40