1、 David Banks is a senior associate at CPP wind engineering and air quality consultants, Fort Collins, CO Some Building Design Issues Related to Extreme Winds David Banks, PhD ASHRAE Member ABSTRACT Pressure from high winds can exceed the pressure provided by HVAC system fans. This can be critical fo
2、r situations where the fans are expected to maintain a pressure difference between one building space and another. One example of this is the smoke control fans used to remove smoke from large open spaces like atria. High winds can reverse flow through these fans. These winds can also cause a “short
3、 circuit”, as air enters through one makeup air location and leaves though another. Makeup air will also enter more quickly than desired, knocking over the smoke plume. For some designs, it is not wise to attempt to operate such a smoke control system during a hurricane. With 100 mph (44 m/s) wind g
4、usts outside, it is unlikely that the air speeds at the makeup openings will be kept below 200 fpm (1 m/s). High winds can significantly affect indoor pressures as well, causing doors to either become difficult to open or difficult to close. Revolving doors are not immune to wind related problem, si
5、nce they are often designed to collapse to allow panicking people to exit in the even of a fire. This “bookfold” mechanism has been triggered by high winds, leading the large glass doors to suddenly collapse, certainly a counterproductive aspect of a safety-related design feature. Wind gusts are cri
6、tical in situations like these, so it is important to understand the fundamentally unstable nature of wind flow around buildings. CLADDING DESIGN PRESSURES The use of boundary layer wind tunnel testing to measure wind pressures is common for structural engineering design. Occasionally, the results o
7、f one of these studies ends up in the hands of the mechanical engineer, in which case it is important to understand how these pressure differ from those typically calculated using procedures outlined in ASHRAE handbooks. For example, I once received a phone call from an engineer who was designing th
8、e louvers through which makeup air would flow into the building in the event that the smoke removal fans were activated. He had been given a cladding pressure diagram that I had produced as part of a wind tunnel study for the structural engineer. The inward-acting wind pressure prescribed in the dia
9、gram was over 30 psf, or 6 inches of water. “It is too much pressure for the louver motors to overcome. Are you sure this pressure is correct?” he asked. In the climate where this building was to be constructed, the design winds for structural purposes are generated by hurricanes, which is true of a
10、ny location near the US east coast or gulf coast. The short answer to his question is that 30 psf (nearly 1500 Pa) is not an especially high value for cladding pressure in a hurricane climate. ASCE 7, the Standard used by building codes across the US, actually specifies that a minimum value of 10 ps
11、f (nearly 500 Pa) be used unless a compelling case can be made that a lower value is safe, and pressures above 100 psf (above 5000 Pa) are not unusual near the edges of roofs. Figure 1 illustrates the relationship between wind speed and the pressure across a wall for a typical range of differential
12、pressure coefficient (DCp) values from ASCE 7. The differential pressure is the difference between the external and internal pressure coefficients. LV-11-C032264 ASHRAE Transactions2011. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in AS
13、HRAE Transactions, Volume 117, Part 1. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAES prior written permission.0246810121020304050wind Speed (m/s)WindPressure(inchesofwater)01020304050600 2040608010wind s
14、peed (mph)Windpressure(psf)DCp= 1DCp= 2DCp= 3Figure 1 Wind pressure vs wind speed for various differential pressure coefficient values. 1 inch of water = 5.2 psf = 249 Pa. WIND PRESSURE AND SMOKE CONTROL The big-picture answer to his question is that louvers that are unable to operate in such high w
15、inds might be a beneficial design feature. Atrium smoke control systems are designed to remove hot, soot-filled air from the smoke layer that collects near the ceiling of the large space during a fire, thus keeping the lower levels of the atrium or mall free of smoke so that they can be used for egr
16、ess during the building evacuation. NFPA 92b, the standard used for the design of smoke control systems by building codes across the US, indicates that makeup air speeds should be limited to 200 fpm (1 m/s) to avoid knocking over the smoke plume, because if the plume is knocked over the lower levels
17、 can fill with swirling smoke, and the lower level egress routes are no longer tenable. The 1 m/s value is not always strictly necessary, as performance-based analyses and experimental measurements can show in specific situations that higher velocity makeup air is not significantly detrimental. Howe
18、ver, as makeup air speeds approach the speed of the hot rising air in the plume, the air flow pattern in the atrium will become dominated by the jets of makeup air rather than the buoyancy of the smoke, and swirling of the plume is inevitable. This will happen at lower air speeds for smaller fires.
19、In the absence of any wind, the makeup air speed is dictated by the makeup air opening area and the exhaust rate. If the exhaust rate is 100,000 cfm (50 m3/s), then the rule of thumb is that 500 sq ft (50 m2) of free area is needed for the makeup air. Of course, this ignores the reduction in effecti
20、ve area as a result of discharge coefficient, but then the makeup air velocity restriction is a rough estimate anyway. At 100 mph (44 m/s), the pressure difference created by the wind between the smoke exhaust above the roof and a makeup air inlet can be 5-10 inches of water (1.2-2.5 kPa). This coul
21、d be enough to reverse flow through the fans and short-circuit winds between makeup air openings on different faces of the building. In addition, wind pressures across the face of a building fluctuate in both time and space, and these fluctuations happen more quickly as wind speed increases. Even op
22、enings on a single faces may see intermittent inward and outward flow in such winds. 2011 ASHRAE 265So there is no reason to expect the smoke flow patterns in such winds to resemble the orderly axisymmetric plume and stable smoke layer scenario envisioned in the design. However, it is not obvious th
23、at the atrium would be safer if the system did not operate, particularly if there is not a large reservoir for smoke collection. If the makeup air louvers open, then the atrium will see tremendous airflow, so the smoke will certainly be removed quickly, even if much of it is removed out of the makeu
24、p air openings. NFPA 92B acknowledge the importance of winds, as it stipulates in section 4.8 that “Designs shall incorporate the effect of outdoor temperature and wind on the performance of the smoke management system”, adding in the explanatory material that the 1 percent extreme wind velocity fro
25、m ASHRAE Fundamentals Climactic Design Information chapter (Chapter 14 in 2009) is suggested as a design condition (section A.4.8). However, NFPA 92B does not suggest how to estimate wind pressure. ASHRAEs Principals of Smoke Management (Klote and Milke, 2002) does provide a method, using mean press
26、ure coefficients. These 1% wind speed values are typically close to 10 m/s (22 mph), and the external mean pressure coefficients are generally less than 1, so that pressures are around 0.2 inches of water. Neither document suggests how the smoke management system might deal with wind pressures, thou
27、gh. At 0.2 inches of water, the wind is unlikely to reverse flow in the fans. It might give us pause in considering the flow rate through the fans, as could be enough pressure to noticeably change the system curve/ fan curve intersection. To estimate this accurately, I suggest following the method d
28、escribed by Cook (1999) to calculate internal pressure based on the pressure at all of the openings. Once the internal pressure has been determined, we can estimate the flow speed through each orifice in a situation dominated by the wind pressures. Typically, the air speed through the makeup air ope
29、nings is 30-70% of the reference wind speed. This would be 10-30 m/s (2000-6000 fpm) during the hurricanes peak winds, and 3-7 m/s (600-1400 fpm) during the 1% winds. 50 YEAR VS. 1% The large difference (a factor of 20 to 50) between the ASCE and ASHRAE design wind pressures is the result of three f
30、actors: 1. The return period, as shown in Figure 2. A wind speed that is exceeded 1% of the time has a return period of about 4 days. Once safety factors are considered, the ASCE design wind pressure has a return period of 300-1700 years, depending on the importance of the building. 2. Averaging tim
31、e has a significant effect on speed. The ASCE design wind speed is a 3 second gust, while the ASHRAE values are generally assumed to be 1 hour averages. The relationship between averaging duration and wind speed for 10 m in open country is commonly derived from the “Durst curve”, adapted in Figure 3
32、 from a plot provided in the commentary to ASCE 7-05. Wieringa (1972) derived the following formula allowing gust factor to be calculated for a range of turbulence levels: G = 1+0.42ln(3600/t)I (1) where G is the gust factor (peak wind speed / hourly mean speed), I is the turbulence intensity (stand
33、ard deviation of wind speed/ mean speed), and t is the averaging time (in seconds). One thing that both the ASCE design wind speed and the wind speeds tabulated in ASHRAE Fundamentals have in common is that they represent winds at a height of 10 m (33 ft) in open, unobstructed terrain. This is also
34、the case when the cable TV weatherman says that a hurricane has “maximum sustained winds of 100 mph”; this means that the peak one-minute-average wind speed at 10 m in open country is 100 mph, or 44 m/s. 3. ASHRAE wind pressure calculations are typically done using mean pressure coefficients, which
35、are 2-3 times smaller than the peak pressure coefficients used for ASCE wind loads. 266 ASHRAE TransactionsGumbel or Type 1 fitA = 9.2, Uo=54 mph80.090.0100.0110.0120.0130.00 200 400 600 800 1000 1200 1400N: wind speed exceeded once in N years3 secondgustwindspeed, mphFigure 2 3-second gust wind spe
36、ed vs. recurrence interval. 100 mph = 44 m/s. 1.001.101.201.301.401.501.601.701.801 10 100 1000 10000gust duration (seconds)Gustfactor=U/U(mean for1hour)I = 0.17 , WieringaI = 0.2 , WieringaDurst, as presented in ASCE-7 commentaryFigure 3 The effect of averaging time on peak wind speed. Would the lo
37、uvers have opened? The design pressure in the cladding diagram did not include a safety factor, and it was an estimate of the peak pressure that the cladding will see during a 50 year wind event, i.e. a wind speed with an annual probability of exceedance of 0.02. This peak wind pressure event can ha
38、ve a very short duration. The gust that pulls cladding from a wall or lifts a ballasted solar panel from a roof might last a tenth of a second or less. From a structural perspective, this is a failure that must be avoided, which is why designers have been required to add a safety factor to this pres
39、sure which effectively means that a 2011 ASHRAE 267failure is not expected until the once-in-700 year wind gust comes along. Figure 2 shows a sample relationship between wind speed and recurrence interval for a region of the US where thunderstorms determine the design wind speeds (i.e. no hurricanes
40、). As an aside, there is often considerable debate (often between homeowners and insurance companies) regarding exactly how a particular building came to be destroyed during a hurricane. Was it due to flood waters (not typically insured) or wind (covered under most policies)? Data showing that the o
41、nce in 50-year design wind speed was exceeded during the storm is often used to back up claims that it was the wind. However, if the building was built using appropriate values from the ASCE-7, the first signs of structural failure should not be seen until the 720 year wind speed is exceeded. After
42、looking at the specific probabilities for the louver motors, analysis showed that during some 10 20 seconds out of the next 50 years, the louvers would not open immediately, but would fight the wind for a few seconds before opening. The odds that the design fire coincides with these 10-20 seconds ar
43、e very small. In addition, a delay of a few seconds would not prevent the smoke evacuation system from working as designed. So the cladding design pressure is not an appropriate design speed for use in the louver-motor design. Cladding pressures can be the appropriate pressure to use for revolving d
44、oors, however. DOORS AND THE WIND Wind causes three types of operability problems for end-hinged doors: 1) Positive (inward-acting) wind pressures make doors difficult to open, 2) Negative (outward-acting) wind pressures prevent doors from closing and latching, and 3) High winds catch the open doors
45、 and swing them around. This last problem can often be addressed by shielding the doors from high winds using fences or trees. The first two problems, however, are the result of the pressure pattern created by the building. A large stagnation zone forms on the upwind face, including the outside surf
46、ace of any doors. The other surfaces of the building will see negative pressures, especially the roof. The pressure across the doors is the difference between the external wind pressure and the pressure in the lobby, which is usually controlled by the HVAC system. Often, the HVAC system is designed
47、to create a mild positive pressure in the lobby relative to the reference pressure at the air intake location. However, even if the wind pressure at the air handlers is not enough to significantly shift the operating point of the fans, the HVAC system may be the dominant leakage path into the lobby,
48、 and the pressure in the lobby will track changes at the HVAC intake or exhaust (with an offset due to the fan pressure). Note that the internal pressure will track pressure fluctuations on the roof, so the mean pressure may not always be an adequate predictor of the portion of the problem that coul
49、d be due to internal pressure, particularly if there are only 1 or 2 intakes. When there are many leakage paths, however, the loss of correlation between the wind pressure fluctuations at the intake locations means that the mean pressure is suitable. Note that a door open to the outside will immediately control the pressure in the lobby. If the air handlers are taking in air from the roof, where the wind pressure is generally negative, and the doors face upwind, leakage will be inward at these doors, and hinged doors will be difficult to open often enough that people will complai
