You are on page 1of 16

ARTICLE IN PRESS

Journal of Wind Engineering and Industrial Aerodynamics 93 (2005) 873888 www.elsevier.com/locate/jweia

Wind effects of parapets on low buildings: Part 4. Mitigation of corner loads with alternative geometries
Gregory A. Kopp, Christian Mans, David Surry
Alan G. Davenport Wind Engineering Group, Boundary Layer Wind Tunnel Laboratory, University of Western Ontario, London, Ont., Canada N6A 5B9 Received 14 May 2004; received in revised form 23 August 2005; accepted 26 August 2005 Available online 10 October 2005

Abstract This is the fourth paper in a series on the wind effects of parapets on low-rise buildings. This part focuses on alternative parapet geometries which can mitigate local (component and cladding) loading due to the formation of corner vortices. It was found that spoilers and porous perimetric parapets signicantly reduce the loads for all areas in the corner, edge and interior zones considered. For solid parapets, raising the corner, or putting a slot in the corner are also benecial when compared to uniform, continuous parapets. Removing the corner of a continuous parapet also lowers the loads as compared to isolated (single) parapets which end at the side wall, indicating that shortened isolated parapets could also be benecial. r 2005 Elsevier Ltd. All rights reserved.
Keywords: Wind loads; Low-rise buildings; Building codes; Parapets

1. Introduction 1.1. Background Worst suction coefcients on the roofs of low buildings are known to occur in the corner, caused by the development of conical vortices during cornering winds [1,2]. These
Corresponding author. Tel.: +519 661 3338; fax: +519 661 3339.

E-mail address: gakopp@uwo.ca (G.A. Kopp). 0167-6105/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.jweia.2005.08.004

ARTICLE IN PRESS
874 G.A. Kopp et al. / J. Wind Eng. Ind. Aerodyn. 93 (2005) 873888

vortices are triggered at the leading corner of the building, and have resulted in worst suction coefcients of over 20 (referenced to the mean wind speed at roof height) being measured [3]. Several researchers have examined different strategies for mitigating these high suctions. As described by Surry and Lin [4,5], there are several types of geometric congurations that can be used to alleviate the corner suction pressures, which they categorized as: 1. Displace the corner vortices using a solid parapet to raise the vortices away from the roof surface. 2. Disrupt the formation of the vortices using a partial or porous parapet. 3. Disturb the vortices on the rooftop by placing a fence or object along the path of formation. 4. Eliminate the straight sharp edges that create the separated ow, e.g. rounded edges. Each of the above methods have been previously studied in wind tunnel experiments [1,413], with a detailed listing of the congurations given in Table 1. These published results are discussed in detail below. 1.2. Solid parapets The solid parapet is the more popular of the congurations, primarily due to its large architectural appeal. The solid parapet acts to raise corner vortices off the surface of the roof so that the intense shear layers no longer interact with the roof surface. A detailed study on the effect of a solid perimetric parapet was performed in Part 1 of the current project [13]. The investigation concluded that parapets with heights, h=H h40:23, may reduce the corner and edge suction pressures by over 50%. However, it should also be recalled that taller parapets also create substantial structural loading on the interior frames and roof regions and severe positive (downward) loading upstream of the base of the leeward parapet, as well as increased loading on the parapet itself. 1.3. Partial and porous parapets A number of partial parapet congurations have been previously tested [5,7,8], differing in geometry by either increasing or decreasing the parapet height in the corner region. These congurations included: (a) discontinued parapet at the corner [7,8], (b) increased parapet height at the corner [7,8], (c) slots, allowing venting, placed in the lower half of the parapet in the corner [7,8], (d) a partial parapet placed a small distance away from the roof corner [4,5], and (e) a castellated parapet [6]. The results from Stathopoulos and Baskaran [7,8] suggest that, in most cases, the partial parapets caused a reduction in loading compared to the solid parapet, but failed to reduce the corner suction pressures compared to the no parapet case. Similar results were observed in [6] for a castellated parapet, with the exception of 1.5 m (5 ft) high parapets, which were found to reduce the suction coefcients in the corner region. However, for this parapet height (1.5 m, with an eaves height of 96 m), the higher perimetric parapet may be equally effective. It should also be noted that the study by Stathopoulos and Baskaran [4,5] was performed on a tall building H 96 m and as such the results may not be the same as for low-rise buildings.

Table 1 Summary of previously published test parameters


Wind angles Terrain Model scale Geometry

Author

Plan dimensions

Eaves height 501


Spoiler (perimetric)
10
6.0" 9.0"

Banks [1] & Wu [14]

12:2 9:1 m; 1:60 gable roof slope open 1:25

4.0 m

Porous parapets (perimetric) (i) wire mesh, 50% solid (ii) 50% solid with circular holes (iii) slatted fence

2.0" overhang

Stathopoulos & Baskaran [7,8]

61:0 61:0 m; at roof 0, 30, 45, 60, 901 open & urban 1:400

4.6 m

Partial parapets, h = 1.0 m solid

slotted

ARTICLE IN PRESS

G.A. Kopp et al. / J. Wind Eng. Ind. Aerodyn. 93 (2005) 873888

discontinuous

raised

875

876

Table 1 (continued ) Wind angles Terrain Model scale Geometry

Author

Plan dimensions

Eaves height 0901 (13 angles)


Sawtooth parapet Cylinders on roof Radial splitters

Surry & Lin [4,5]

12:2 9:1 m; 1:60 gable roof slope open 1:50

4.0 m

Partial parapet

50% solid (h=3.0m)

ARTICLE IN PRESS

Curved eaves

Robertson [10]

24:1 12:8 m; 101 pitch, gable roof N/A open

5.3 m

Full-scale

R=2.1"

G.A. Kopp et al. / J. Wind Eng. Ind. Aerodyn. 93 (2005) 873888

Blackmore [9]

75:0 75:0 m; at roof 0, 15, 30, 45, 60, 75, 901 open

25.0 m

1:250

Slope=30, 45, 60

w=20.5', 41.0'

ARTICLE IN PRESS
G.A. Kopp et al. / J. Wind Eng. Ind. Aerodyn. 93 (2005) 873888 877

Conguration (d), tested by Lin and Surry [4,5] with a building height of 4 m, was found to reduce the peak and rms corner pressures by about 25%, but caused no signicant change in the mean corner pressures. The differing conclusions between the studies [46] can probably be attributed to the differences in building height, in addition to variations in the locations of the pressure taps. Baskaran and Stathopoulos [7,8] recorded a total of 87 pressure taps over the entire roof surface, however, the majority of the presented results were from a single corner tap. The study by Lin and Surry [4] recorded pressures from 30 taps at various locations around the building corner. 1.4. Fences and spoilers Lin and Surry [4,5] examined three saw-tooth congurations (single, dual or triple crowned saw-tooth parapets) positioned either at the leading corner or a short distance away from the corner. The saw-tooth conguration reduced the peak corner pressures by 3040%, with the dual or triple saw-tooth parapets yielding no further reductions over the single saw-tooth conguration. The largest reductions in loading occurred when the parapet was positioned ush with the leading corner of the building. The porous parapet has consistently been found to be the most effective parapet conguration, reducing the corner pressures by up to 70% [5]. As described by Surry and Lin, the effectiveness of the porous parapet is due to three factors: (a) the porous screen disrupts the vortex formation along the roof edge, (b) the screen absorbs some of the energy of the ow over the roof edge, and (c) the weakened vortices that do form, do so over the top edge of the parapet and are displaced away from the roof-top. Recent work by Banks [1], who examined a number of porous parapet geometries, found similar results. Banks studied three variations: a mesh of 50% porosity, a circular hole porous parapet with 50% porosity and a slatted fence. The effectiveness of the three parapet congurations were similar (around 50% reduction in peak pressures), with the circular holed parapet providing marginally better results. Banks [1] and Wu [14], in a collaborative project between Colorado State University (CSU) and Texas Tech University (TTU), also created a novel device, a spoiler, specically designed to be easily added to existing structures. (See Table 1 for details of its appearance.) This spoiler is relatively small in full-scale, 10 cm by 2 cm, but was still found to reduce the peak coefcients by about 50% in wind tunnel testing at CSU and full-scale testing at TTU. It does this by allowing air to pass beneath the device, disrupting the formation of vortices at the sharp corners and edges. Melbourne [15] used similar principles to alleviate the high suctions on the leading edges of cantilever stadium roofs. Roof-top splitters, either solid or porous, have been positioned along the approximate lines of the corner vortices in a direct effort to interfere with the formation and development of the vortex. Lin and Surry [4,5] found that this is an effective method of eliminating the formation of the vortices, reducing the corner pressures by 60%. 1.5. Curved and chamfered eaves edges The effect of rounding the eaves of industrial buildings is known to signicantly reduce the worst suction pressures at the leading edge. Full-scale studies by Robertson [10] recorded a reduction of about 40% in the mean pressure coefcients along a line of taps at the mid-span (i.e., L/2) of a low-rise building with a curved edge at the eaves. However,

ARTICLE IN PRESS
878 G.A. Kopp et al. / J. Wind Eng. Ind. Aerodyn. 93 (2005) 873888

wind loads behind the ridge of the building are increased by about 30%, possibly because there is less energy lost at the building edge resulting in higher velocities at the ridge. Unfortunately, no results are presented for winds approaching the leading corner of the building where the worst suction pressures are expected to occur. However, Lin and Surry [5] recorded the effect of rounded edge proles on the corner pressures, reporting a reduction of about 60% in the peak and mean pressure coefcients. Work by Cooper [16] on rounding the leading edges of truck trailers provides a useful guide to minimum radius requirements, although this should probably be repeated for low buildings in thick turbulent boundary layers. Similar to the rounded edge prole, Blackmore [9] examined the effect of chamfering the edge prole of a building as a means of preventing the formation of the corner vortices. Six alternative congurations were examined: three chamfer angles, 30o, 45o and 60o, with two chamfer widths. Results suggest the largest load reductions of 70% were observed for the 30o chamfer, which was the least steep of the three tested. 1.6. Objectives of the current study The above considerations suggest that the porous parapet may be the most effective parapet geometry for mitigating corner suction pressures. Independent experiments by Lin and Surry [5] and Banks [1] concluded that porous parapets, with about 50% porosity, reduced the worst loading by around 5070% in the corner region. Saw-tooth parapets were less effective, reducing the corner pressures by about 3040%. Solid and partial parapets were the least useful, reducing the loading only if the height of the parapet was greater than 1.5 m (5 ft) for the building heights they used. When reviewing the literature we noted that many of the previous experiments were performed with a limited number of pressure taps and that little area-averaged information, the type of loading information designers require, was available. The high-resolution module that we used in Part 1 [13] allows detailed area-averaged loads to be obtained without the inherent uncertainty of relying on single point pressures or distributions of point pressures. Thus, the objective of the current work is to determine the area-averaged (component and cladding) loads in a corner panel using alternative parapet geometries which have been shown to work, but which would be possible and practical for industry to implement. 2. Experimental set-up Testing was conducted in the Boundary Layer Wind Tunnel II at the University of Western Ontario. The 1:50 scale model used in Parts 1 [13] and 3 [17] of this study, was used for the current experiments. The building model has equivalent full-scale plan dimensions of 31.1 m (102 ft) by 46.3 m (152 ft) with a 1 4 on 12 gable roof slope. Testing was repeated for the eight congurations listed in Table 2. These were chosen in consultation with Dr. Lee Shoemaker of the Metal Building Manufacturers Association as to which types may possibly be used in industry. Note also that the isolated parapet (i.e., a parapet along a single wall) data, reported in Part 1 [13], are also used herein. About 700 pressure taps were used on the high resolution module, shown in Fig. 1(a). The high-resolution module was placed in the corner (see Fig. 1b) with its edge the thickness of the parapet (0.3 m) from the outside of the wall, as in Part 1 [13]. Photographs of the building model with the slotted parapet and the spoiler are given in Fig. 2.

ARTICLE IN PRESS
G.A. Kopp et al. / J. Wind Eng. Ind. Aerodyn. 93 (2005) 873888 Table 2 Congurations in the present experiments Plan dimensions Eaves height Model scale Terrain Wind angles Geometry of parapets 31:1 46:3 m; 1:48 gable roof slope 4.6 m 1:50 Open zo 0:03 m 3251 1. no parapet 2. solid, h 0:9 m, 0.30 m thick, perimetric parapet 3. 50% solid screen, h 0:9 m, perimetric parapet 4. slotted parapet, h 0:9 m 879

5. solid parapet ending h from side wall (no corner)

6. as in 5, with 50% solid screen corner (porous corner) 7. solid parapet with raised corner

8. perimetric spoiler

Measurements were obtained in a single terrain condition, an open country terrain zo 0:03 m for a single wind angle (3251), consistent with the worst wind direction for corner vortices for buildings with solid perimetric parapets of height h=H ho0:17 [13]. The open country terrain was identical to that in [13,17] (although the current data were obtained 3 years later than the earliest data reported in [13]).

ARTICLE IN PRESS
880 G.A. Kopp et al. / J. Wind Eng. Ind. Aerodyn. 93 (2005) 873888

0.076m

7.6 m 0.076 m 47 @ 0.152 m 0.38 m X EDGE

CORNER 3.7 m 21 @ 0.152 m

EDGE

INTERIOR

(a)

0.38m Y L X 90

46.3 m

W 0 Corner 7.6m Edge

31.1 m

180

Y 3.7m 13.4m

(b)

270

Fig. 1. (a) pressure tap layout for the high-resolution module, and (b) over model dimensions and denition of the wind angle. Note that only the corner location of the high-resolution model is tested in this part.

Pressure measurements were made at a wind tunnel reference speed of 14 m/s. This leads to an eaves height speed of 8.6 m/s in the open country terrain. The reference location was at the mid-height of the wind tunnel, just upstream of the model. Each pressure tap was sampled essentially simultaneously for 120 s at a rate of 400 samples per second and

ARTICLE IN PRESS
G.A. Kopp et al. / J. Wind Eng. Ind. Aerodyn. 93 (2005) 873888 881

Fig. 2. (a) Close-up photographs of the high-resolution module with the raised parapet in the corner. Machine nuts on the oor of the wind tunnel provide roughness right up to the edge of the model. Tubing can be seen through the clear acrylic roof below the high-resolution module. (b) Photograph of the spoiler at one corner of the model.

digitally low-pass ltered at 200 Hz. The measurements recorded within the sampling cycle have a maximum lag time of about 15/16 of the sampling rate, which is 15=16 0:002 1:875 ms. This time lag is corrected by linear interpolation of the data within the same sample cycle. The resulting pressure time series were then referenced to the mean dynamic velocity pressure at eaves height, H, using the previously measured velocity prole, as in [13]. The peak pressure and area-averaged load coefcients presented in the study are not the absolute worst coefcients recorded within the sample time, but are Lieblein-tted statistical peaks. This involves dividing the recorded time series into ten equal segments and performing the Lieblein BLUE formulation [18] with the peak values taken from each of 10 segments. The resulting mode and dispersion of the Type I extreme value distribution were used to determine the mean 1 2 hour (full scale) peak value for each pressure and load coefcient reported herein. These are believed to be more statistically stable quantities, than the actual recorded peaks.

ARTICLE IN PRESS
882 G.A. Kopp et al. / J. Wind Eng. Ind. Aerodyn. 93 (2005) 873888

-2.0 -1.6 -1.2 Cp -0.8 -0.4

0.7 0.6 0.5 Cp' 0.6 x/H 0.4 0.3 0.2 0.1

0.0 0.0 (a) -6.0 -5.0 -4.0 Cpmin -3.0 -2.0 -1.0 0.0 0.0 (c)

0.0 0.3 0.9 1.2 1.5 (b) 20.0 16.0 12.0 g 8.0 4.0 0.0 0.3 0.6 x/H no parapet spoiler 0.9 1.2 1.5 (d) slotted corner raised corner 0.0 0.3 0.6 x/H porous corner no corner 0.9 1.2 1.5 0.0 0.3 0.6 x/H 0.9 1.2 1.5

0.9m parapet (perimetric)

porous perimeter
Fig. 3. (a) mean, (b) rms, (c) minimum pressure coefcients, and (d) peak factor distributions along the line y=H 0:42 for a wind angle of 3251 for the various parapet geometries.

3. Results and discussion 3.1. Distribution of point pressures Fig. 3 depicts the distribution of the mean, Cp, and root-mean-square1 (rms), Cp0 , point pressure coefcients along the line of taps nearest the boundary between the ASCE 7-02 [19] dened corner, edge and interior zones (see Fig. 1a), i.e., y=H 0:42, for the eight test congurations. Perhaps the most striking observation that can be made is that the different parapet geometries lead to mean Cp distributions which can be categorized into two groups. One group has mean distributions which cluster around the data from the solid,
Root-mean-square (rms) is used as a synonym for standard deviation in the current work, i.e., it is always calculated with the mean component removed.
1

ARTICLE IN PRESS
G.A. Kopp et al. / J. Wind Eng. Ind. Aerodyn. 93 (2005) 873888 883

perimetric parapet with parapet height, h 0:9 m. This group consists of parapets which can be viewed as perturbations to the solid, perimetric parapet case, namely, the parapets with slotted corners, porous corners, raised corners and no corner. It is clear that this group would not be expected to signicantly alter the large-scale wind loading, although particular features of the local loading near the corner are clearly affected. This will be examined in detail in the following section. The second group is dened by congurations having common mean pressure distributions which are clustered away from those in the rst group. The group contains the no parapet, the spoiler and the porous, perimetric parapet cases. This second group clearly has much lower mean loads associated with it, at least for this particular line of taps. (It should be noted that no attempt, either here or in Banks study, has been made to measure the loads on the spoiler itself, but it is likely to be attracting high lift loads as it re-directs the corner ow, so that its attachment to the roof needs to be equally considered.) The categories of behaviour dened by the mean pressure distributions in Fig. 3(a) also appear to hold for the rms pressure coefcients, Cp0 , distributions presented in Fig. 3(b). However, within these two groups, differences in the resulting corner ow become apparent. The group of ve parapets that are perturbations of the solid parapet, the rst group described above, exhibits interesting behaviour. Clearly, the worst conguration is the solid, perimetric parapet which has elevated rms levels due to a strong and stable vortex core [13] and its expanded aerial extent on the roof surface. The three geometries with openings in the corner of the parapet (the slotted corner, porous corner and no corner) all exhibit similar rms distributions which are consistent with the presence of corner vortices. The resulting surface pressures are smaller in magnitude than for the solid parapet, but still larger than the no parapet case within 1H of the edge. The raised corner also gives indications that the corner vortex remains, but that it is raised higher from the surface, much like a taller solid parapet (e.g., see Fig. 6(b) in Part 1). One should note though that rms levels are signicantly higher for the raised parapet, so that one would expect higher area-averaged loads for this conguration, all else being equal. All of these geometries then lead to worse peak pressures than the non-parapet case, but they are better than the solid perimetric parapet case, at least for this line of taps. Thus, it appears that corner treatments to these parapet congurations will ease the local loading if a parapet is required, but that for h 0:9 m on a building with H 4:6 m (h=H h 0:17), this is worse than having no parapet. However, analysis of area-averaged loads will be needed to conrm this. Finally, it is interesting to note that beyond x=H 1, Cp and Cp0 asymptote to the same values for all ve congurations in the group, as does Cp0 for h 0. Thus, the effects of the perturbations to the parapet geometry are truly local, as one would expect. For the second group of parapets dened above (i.e., the spoiler, porous perimetric parapet and no parapet), the classication based on the mean distribution still holds. Fig. 3 clearly indicates that the corner vortices are signicantly weakened so that overall loads are much lower than even the no parapet case, although the typical footprints of the corner vortices are still observed. Fig. 3(c), which depicts the minimum pressure coefcients, shows the same trends. Also, the minima are roughly bounded by the no parapet and solid parapet data (except for the spoiler and the porous parapet), with a few exceptions. Not surprisingly, the boundary between the two groups is still clear, although it would appear that the no parapet case ts much more with the rst group. What all of this implies

ARTICLE IN PRESS
884 G.A. Kopp et al. / J. Wind Eng. Ind. Aerodyn. 93 (2005) 873888

is that the peak factor g Cpmin Cp Cp0 (1)

is strongly affected by the parapet geometry. This is shown in Fig. 3(d). Corner vortices (for h 0) have a large quiescent zone between them, as can be seen by the low mean suction for 0:2ox=H o0:6 in Fig. 3(a), with the smallest rms value at x=H 0:3. The pressure gradients due to the corner vortices are clearly large. Since the vortex cores will move randomly about a well-dened mean location [1], large suctions will occur in the quiescent region from time to time due to a vortex core moving slightly or being strengthened from a large gust. This behaviour, with different ow phenomena passing a particular point, results in large peak factors (up to 18!) in Fig. 3(d). Solid perimetric parapets stabilize this phenomenon, most likely because, as the core is further from the roof surface, the pressure gradient is reduced and a slight movement of the core has relatively less effect. In the range 0:2ox=H o0:6, the peak factor is in the range of 4 to 5 for h 0:9 m. These values, together with the raised corner, are the lowest observed along this line. 3.2. Local, area-averaged loads Within each zone, a series of area-averaged loading coefcients were developed by simultaneously combining pressures from a number of surrounding taps. The Lieblein method, discussed above, was then applied to obtain statistically reliable peak values from the time series of the area-averaged coefcients. For simplication, a square-shaped geometry was consistently used in the analysis, ranging in full-scale area from about 0.1 to 5 m2, as listed in Table 3 of [13]. A few rectangular areas were considered for the larger areas as well. The roof surface was separated into three regions (corner, edge and interior zones), as dened by the current ASCE 7-02 standard. For the tested building geometry, this corresponds to an edge zone width of 1.8 m (6 ft). Dimensional areas will be discussed, since this is the format currently applied in building codes such as the ASCE 7-02, even though normalizing areas by H2 has been demonstrated to collapse such data, as discussed in Part 1. The coefcients obtained from this method have been converted to effective GCp values for direct comparison with the coefcients used in the ASCE 7-02. St. Pierre et al. [20] have described in detail the procedure we are using for comparing wind tunnel and code loads. In particular, referring the wind tunnel coefcients to the ASCE coefcients requires GCpeq ^ qH Cp ^ , F WT Cp q10 m;3 s K zt K h K d I (2)

^ is the peak coefcient where (GCp)eq is the equivalent wind tunnel pressure coefcient, Cp based on the mean (e.g., hourly) wind pressure at the eaves height, qH, q10 m,3 s is the basic wind pressure in the ASCE 7-02, Kzt is the topographic factor, Kh converts the pressure coefcient to roof height (terrain factor), Kd is the directionality factor and I is the importance factor. All factors were set to the same values as used Part 1 [13]. Figs. 46 depict the worst area averages obtained in the ASCE 7-02 dened corner, edge and interior zones, respectively. The coefcients, GCp, in the gures are in a form suitable for direct comparison with the coefcients in ASCE 7. The conversion of the experimental

ARTICLE IN PRESS
G.A. Kopp et al. / J. Wind Eng. Ind. Aerodyn. 93 (2005) 873888 885

-4

-3.2

-2.4 GCp -1.6 -0.8 0 0.01

0.1 Area (m2) no parapet raised parapet slotted corner porous perimeter 0.9 m parapet (isolated)

10

0.9m parapet (perimetric) no corner porous corner spoiler ASCE 7-02 corner

Fig. 4. Variation of worst pressure coefcients with loading area in the corner region of the high-resolution module.

-3.2

-2.4

GCp

-1.6

-0.8

0 0.01

0.1 Area (m2) no parapet raised parapet slotted corner porous perimeter 0.9 m parapet (isolated)

10

0.9m parapet (perimetric) no corner porous corner spoiler ASCE 7-02 edge

Fig. 5. Variation of worst pressure coefcients with loading area in the edge region of the high-resolution module.

coefcients was done in precisely the same manner as by St. Pierre et al. [20], and in Part 1 [13]; the reader is referred there for details. One major difference between Part 1 and the present work is that herein, only one wind angle is used, whereas in Part 1, the worst values

ARTICLE IN PRESS
886 G.A. Kopp et al. / J. Wind Eng. Ind. Aerodyn. 93 (2005) 873888

-2

-1.5

GCp

-1

-0.5

0 0.01

0.1 Area (m2)


no parapet raised parapet slotted corner porous perimeter 0.9 m parapet (isolated)

10

0.9 m parapet (perimetric) no corner porous corner spoiler ASCE 7-02 interior

Fig. 6. Variation of worst pressure coefcients with loading area in the interior region of the high-resolution module.

from all wind angles considered are presented. Since the actual worst values may come from other wind angles, the present gures are primarily useful for comparison purposes between the different parapet congurations. In addition, data from the isolated parapet conguration, as reported in Part 1, has been included to aid the discussion. In the corner region (Fig. 4), it is clear that both the spoiler and porous parapet signicantly reduce the local area-averaged loads compared to the no-parapet conguration, as expected from the point pressures presented in Fig. 3. Thus, these devices are useful in reducing the loads on buildings which would otherwise not have parapets. (The reader is also reminded, from the results in Part 1, that high, solid parapets with h=H hX0:23, also reduce the loads.) For the other group of parapets, the corner perturbations to the solid, perimetric parapet, the results are interesting. In the corner zone, the raised corner is an effective method for reducing the loads, as compared to the solid (uniform) parapet, and, for areas less than 1 m2, perhaps even for buildings without parapets (which would then be similar to a saw-tooth parapet [4,5]). The worst cases are the porous corner and the no corner congurations which are also higher than the solid parapet. This appears to occur because the ow behaves somewhat like it would for the case of an isolated parapet, although the isolated parapet data from [13] indicate that the latter conguration is the worst case. Thus, one could conclude that ending a parapet before the corner reduces the corner zone loads, as compared to continuing it all the way to the corner. Fig. 5 shows the results for the edge zone, as dened by the ASCE 7-02. Once again, the spoiler and porous parapet are seen to yield local loads considerably below all other congurations, including the no-parapet case. For all other congurations, there appears to be real benet to modifying the parapet corner geometry, as all data are well below the

ARTICLE IN PRESS
G.A. Kopp et al. / J. Wind Eng. Ind. Aerodyn. 93 (2005) 873888 887

solid parapet case. For areas less than 0.5 m2, these congurations are even better than the no parapet case, somewhat surprisingly. Once again, the isolated parapet is seen to be the worse case, although for larger areas, the data overlap with the solid parapet data as corner effects are reduced. In the interior zone (Fig. 6), the spoiler has the lowest loads, followed by the porous parapet. The no-parapet case is next, but drops more rapidly with area so that for areas of 2 m2, the same values as the porous parapet are obtained. The group of parapets which have corner geometry modications, all behave similarly in the interior zone. And, once again, the isolated parapet conguration is the worst, but with loads approaching the solid perimetric parapet case for areas 2 m2. 4. Conclusions In the current study, alternative parapet geometries were studied in order to nd effective means for reducing the signicant area-averaged loads associated with corner vortices. The major conclusions resulting from this work are: 1. Spoilers and porous perimetric parapets are effective means of mitigating roof suctions in the corner, edge and interior zones as dened by ASCE 7-02. These result in lower area-averaged component and cladding loads, than not having a parapet at all. 2. Given a choice, raising a solid parapet at the corner will reduce loads in all zones compared to a solid continuous parapet. A parapet with a slotted corner would also be benecial, but less so. 3. Single, isolated parapets (i.e., parapets spanning one complete wall of the building) are the worst conguration. 4. Leaving the corner open is worse than a continuous solid parapet for small areas. However, if an isolated parapet is used, ending it before the side wall will reduce loads compared to an isolated parapet brought all the way to the corner. Acknowledgements This project was initiated by the Metal Building Manufacturers Association and the American Iron and Steel Institute. Their nancial support and continued interest is greatly appreciated. The on-going interest of Dr. Lee Shoemaker is greatly appreciated. One of the authors (G.A. Kopp) gratefully acknowledges the support of the Canada Research Chairs Program. The authors also wish to acknowledge the contribution of Ms. L.M. St. Pierre who performed some of the data analysis. References
[1] D. Banks, The suction induced by conical vortices on low-rise buildings with at roofs, Ph.D. Thesis, Department of Civil Engineering, Colorado State University, Fort Collins, Colorado, 2000. [2] D. Banks, R.N. Meroney, P.P. Sarkar, Z. Zhao, F. Wu, Flow visualization of conical vortices on at roofs with simultaneous surface pressure measurement, J. Wind Eng. Ind. Aerodyn. 84 (2000) 6585. [3] T.C.E. Ho, D. Surry, Factory mutualhigh resolution pressure measurements on roof panels, Boundary Layer Wind Tunnel Laboratory Report BLWT-SS11-2000, University of Western Ontario, London, Ontario, 2000.

ARTICLE IN PRESS
888 G.A. Kopp et al. / J. Wind Eng. Ind. Aerodyn. 93 (2005) 873888 [4] J.X. Lin, D. Surry, Suppressing extreme suction on low buildings by modifying the roof corner geometry, in: Seventh US National Confrerence on Wind Engineering, University of California, Los Angeles, California, 1993, pp. 413422. [5] D. Surry, J.X. Lin, The effect of surroundings and roof corner geometric modications on roof pressures on low-rise buildings, J. Wind Eng. Ind. Aerodyn. 58 (1995) 113138. [6] A.G. Davenport, D. Surry, The pressures on low-rise structures in turbulent wind, in: Canadian Structural Engineering Conference, The Canadian Steel Industries Construction Council, Toronto, Ontario, 1974, pp. 139. [7] A. Baskaran, T. Stathopoulos, Roof corner wind loads and parapet congurations, J. Wind Eng. Ind. Aerodyn. 29 (1988) 7988. [8] T. Stathopoulos, A. Baskaran, Turbulent wind loading on roofs with parapet congurations, Can. J. Civil Eng. 29 (1988) 570578. [9] P.A. Blackmore, Load reduction on at roofsthe effect of edge prole, J. Wind Eng. Ind. Aerodyn. 29 (1988) 8998. [10] A.P. Robertson, Effect of eaves detail on wind pressures over an industrial building, J. Wind Eng. Ind. Aerodyn. 38 (1991) 325333. [11] J.X. Lin, D. Surry, The variation of peak loads with tributary area near corners on at low building roofs, J. Wind Eng. Ind. Aerodyn. 7778 (1998) 185196. [12] S. Pindado, J. Meseguer, Wind tunnel study on the inuence of different parapets on the roof pressure distribution of low-rise buildings, J. Wind Eng. Ind. Aerodyn. 91 (2003) 11331139. [13] G.A. Kopp, C. Mans, D. Surry, Wind effects of parapets on low buildings: part 1. Basic aerodynamics and local loads, J. Wind Eng. Ind. Aerodyn., 2005 (under revision). [14] F. Wu, Full-scale study of conical vortices and their effects near corners, Ph.D. Thesis, Department of Civil Engineering, Texas Tech University, Lubbock, Texas, 2000. [15] W.H. Melbourne, J.C.K. Cheung, Reducing the wind loading on large cantilevered roofs, J. Wind Eng. Ind. Aerodyn. 28 (1988) 401410. [16] K.R. Cooper, The effect of front-edge rounding and rear-edge shaping on the aerodynamic drag of bluff vehicles in ground proximity, SAE Technical Paper 850288, 1985. [17] C. Mans, G.A. Kopp, D. Surry, Wind effects of parapets on low buildings: part 3. Parapet loads, J. Wind Eng. Ind. Aerodyn., 2005 (under revision). [18] J. Lieblein, Efcient methods of extreme value methodology, National Bureau of Standards Report No. NBSIR 74-602, Washington, DC, 1974. [19] ASCE 7-02, Minimum Design Loads for Buildings and Other Structures, American Society of Civil Engineers, Reston, Virginia, 2003. [20] L.St. Pierre, G.A. Kopp, D. Surry, T.C.E. Ho, The UWO contribution to the NIST aerodynamic database for wind loads on low buildings: part 2. Comparison of data with wind load provisions for low buildings, J. Wind Eng. Ind. Aerodyn. 93 (2005) 3159.

You might also like