(Dr. Abdul Majid Ismail)
The idea of integrating natural means of ventilation in high-rise buildings has not been proven to be effective, especially in a hot-humid tropical climate. The lack of control, high external air temperature, low quality external air, the lack of "high-tech" image which can be presented to the client, and generally undesirable locations in urban centres could be the reasons why the possibility of natural ventilation has been neglected. This paper discusses the first part of a preliminary investigation into basic design principles of wind-driven ventilation for tall buildings carried out by the author at the Architectural Science Research and Development, University of Wales.
Most of the present high-rise buildings including housing complexes in Malaysia are designed to rely on technology in order to provide comfortable conditions. Artificial ventilation (air-conditioning) and artificial lighting are normally adopted, although natural wind flow and light are freely available much of the time. Natural ventilation especially wind-driven has been proven to be very successful in domestic low-rise environments. However, for domestic high-rise especially medium and high-cost developments, the great prospect of natural ventilation is being eroded by the extensive adoption of air-conditioning. Ironically, in European cities such as the city of Frankfurt, natural ventilation is encouraged to be used not only in domestic high-rise developments but also in commercial high-rise developments.
In Malaysian hot-humid conditions, natural air flow fulfils the function of helping the hot and sticky skin by encouraging the evaporation process to achieve a certain level of comfort. The relative humidity in Malaysia, which remains high and varies from 55% to almost 99%, requires high air movement indoors, especially when the air temperature is also high, between 22 0C and almost 34 0C. The air that is moved through the exchange of indoor with "fresh" outdoor air can provide cooling, and can also act as a heat carrying medium. These natural phenomena are quite similar in both domestic low-rise and high-rise situations. However, in the case of high-rise buildings, there are other additional factors which may influence the external air flow pattern. Consequently, these will further modify the indoor air flow. The higher wind velocity in the upper atmosphere could be utilised as one of the possible cooling methods for a tall building. This natural phenomenon of gradient wind might be very useful, especially in a tropical climate where the mean surface wind velocity is mild.
The mean surface winds over peninsular Malaysia are generally mild, with the mean speed of about 1.5 m/s, and a maximum speed of less than 8 m/s. The main direction is variable. Appendix A shows the annual percentage frequency of various wind speeds and directions (1971-1990) for Penang, Kuala Lumpur and Johor Baharu The hourly speeds are high during the day, and the calm periods, which vary from 16% to 50%, mostly occur at night. Appendix B shows the summary of the 1993’s and 1994’s hourly wind speeds for Penang, Kuala Lumpur and Johor Baharu. These wind conditions are favourable for the adoption of natural ventilation. Effective indoor air flow could be achieved by fully utilising the available winds.
The movement towards "green architecture" or climatically responsive design, which leads towards "modern regionalism" in Malaysia seems to be progressing slowly compared with developments in Europe. In Malaysia very little attempt has been made to promote passive ventilation in countering the trend of solving all indoor comfort problems by simply adopting mechanical solutions especially in high-rise buildings. An excellent naturally ventilated high-rise domestic building in hot-humid climates could be accomplished by having climatically conscious design qualities as the pre-requisite. A brief review of this subject conducted by the author revealed that the input of natural ventilation in high-rise buildings in Malaysia was minimal and not yet comprehensively developed. However, the review gathered a number of basic design elements which are appropriate to the climatically conscious design and natural ventilation of high-rise buildings in the hot-humid climate of Malaysia. The appropriate design elements are:
This first part of the study focuses only on the external wind pressure distributions and flow patterns created by varying the related design parameters derived from 3.3 such as:
Passive ventilation is a type of ventilation process due to naturally produced pressures due to wind and stack effect. The rate is determined by the geometry and construction details of each building. Therefore, designing building for wind-driven natural ventilation requires some understanding on forms and configurations; treatments of external envelopes; and some special features such as location of service core; atria, open central corridors, open ground floor and external wind scoops. All these features which are under the control of the designer will influence the external wind-flow and consequently determine the effectiveness of indoor air-flow. Forms and configurations include building depth and height relative to the surrounding buildings. Treatments of external envelopes include size and location of openings; solar shading projections and external balconies; and corridors. Natural air flow across a building also requires a minimum depth in order to be effective. According to Awbi (1994), in designing building for effective cross-flow ventilation, the building depth should not be more than 12 m deep. Consequently, with this narrow plan form and large openings, wind pressure will be effective in dissipating heat out of the building. The overall parameters involved in the study of wind flow in and around buildings can be summarised in Figure 1.
Figure 1: Factors affecting wind-driven ventilation
To understand the effect of external building features to the wind flow, laboratory scale-model experiments in a wind tunnel to study the effect of external wind conditions and pressure distribution is required. Prior to the scale-model experiments in the wind tunnel, a review on the relevant existing tall building configurations has been carried out by the author which leads to a selection of building forms and configurations that are essential for the wind tunnel study. However, the nature of the experiments and other limitations resulted in further simplification and generalisation of the studies. Further streamlining and analysis against all the essential parameters of wind tunnel modelling and building aerodynamics were also required. A suitable velocity gradient of 0.28 power law model was selected and simulated in the wind tunnel and used in the pressure and flow visualisation tests. This simulation factor leads to the suitable scale factor of 1:200 and a unit dimension of 10 m. Understanding of the basic building aerodynamics leads to the limitation of the test to only bluff body models with sharp edges, stretching from simple to more complex forms. Two slenderness ratios were selected to represent high-rise 100 m tall or 25-storey buildings and medium rise 50 m tall or 12-storey buildings. All together there are six groups of scale-models adopted in the study (Appendix C):Three dimensional objects such as trees, walls, hills and buildings which are projecting above the earth’s surface are aerodynamically rough and generate a frictional force to the wind flow. Hence, a velocity gradient is created to the wind flowing over the earth’s surface. At gradient height, the wind speed is not affected by the friction of the earth’s surface and the velocity at this height is called the gradient velocity. Below the gradient height the mean wind profile follows the power law which is represented by the expression:
Vz/Vg = (z/zg)a .............(i)
where Vz = Wind velocity at a height z from the ground
Vg = Free stream gradient wind velocity
zg = Gradient height
a is an exponent whose value depends on the roughness of the terrain shown
in Table 1.
Table 1: Typical values of zg and a for three types of terrain
| Types of terrain | zg (m) | a |
| Flat open country | 275 | 0.16 |
| Suburban or towns | 400 | 0.28 |
| Heavily built-up areas, city centres | 500 | 0.4 |
Source: Hutcheon, 1983
In this study, ventilation rates are determined from model tests in wind tunnel by measuring the external pressure distribution and using this as data for a theoretical prediction. One important theory of wind tunnel modelling states that the dimensionless ratio, the pressure coefficients (Cp), of a building with sharp edges is independent of the wind speed. This relationship justifies using the pressure measurements on reduced scale-models in the wind tunnel for determining the actual wind pressure on the building. The equivalent pressure due to wind at the building surface in Malaysian conditions is derived from the following:
Figure 1: Velocity profiles over terrain simulated in the wind tunnel
The Malaysian condition:
It is assumed that winds are blowing from the open Malaysian meteorological site of terrain roughness a = 0.16 to a town area of roughness a = 0.28. At the meteorological site, the a = 0.16 power law is appropriate, and the speed at gradient height of 275 m will exceed the 10 m (standard meteorological height) speed by a factor calculated as:
Vg /V10 = {275/10}0.16
Therefore Vg = 1.7 V10 ---------------------(ii)
In the wind tunnel simulation:
A 0.28 power law profile for a town area was simulated in the wind tunnel. Using a similar power law relationship, the free wind speed at height (H) in the tunnel VH is given by:
VH / Vg = {H / 400}0.28
Therefore VH = Vg {H / 400}0.28 -------------(iii)
The speed at gradient height is equal for both power laws, substituting the value of Vg from equation (ii) into equation (iii):
VH = 1.7 V10 (H / 400) 0.28 --------------(iv)
For this part depth simulation, H is the reference height where the reference dynamic pressure (pref) is associated with the reference velocity (Vref).
Therefore, in this case:
VH = Vref.
The appropriate values for computing the ventilation rates are the mean pressure (time-average) values. The time-average surface pressure is proportional to the wind velocity pressure (pw) in Pa given by Bernoulli’s equation:
pw = Cp x 0.5r Vref2 --------------(v)
Where Cp = pressure coefficient
r = density of air (1.225 kg/m3)
The pressure coefficient (Cp ) is determined by the wind tunnel test as:
Cp = p1 /pref
where p1 = surface pressure on the model over the local outdoor atmospheric
pressure, obtained from the pressure tapping.
pref = reference free wind pressure at height H in the wind tunnel.
Therefore, the actual wind pressure on the building in Malaysia (pwm ) can be calculated from the above and is expressed as:
Equation (vi) is used to determine the equivalent pressure due to wind at the building surface in Malaysian conditions relative to the reference velocity derived from the data of the mean wind speed obtained from the Malaysian meteorological stations.pwm = Cp x 0.613 Vref2 --------------(vi)
The effects of wind pressure in the creation of natural internal air flow in high-rise buildings require a number of investigations. In a hot-humid climate such as that of Malaysia, faster air is required to provide cooling of the skin. According to Koenigsberger et al. (1976), the only natural force that can be relied on to provide the required cooling is the dynamic effect of wind. The available wind should be trapped and fully utilised to provide a suitable indoor air flow. The negative effects of too much wind can be controlled manually or automatically by properly designed openable shutters.
Theoretically, air flow through a building is created by a pressure difference between two opposing sides. The pressure difference between any two points on the building envelope determines the potential driving force for ventilation where openings are provided at these points. When wind blows and impinges on a building, a distribution of static pressure is developed over the building’s exterior surface that is determined by the wind direction. According to BS 5925 (1991), there are three major determinants that will influence the distribution of pressure at the surface of the building:The ventilation rate for a given wind speed and direction is estimated when the following information has been established:
Considering only large openings such as windows and doors, the flow rate can be calculated from the following relationship:
Q = Cd A (2dp/r )0.5 (m3/s) ---------------------(vii)
where Cd = discharge coefficient, 0.61(sharp orifice flow)
A = effective area of openings (m2)
dp = (pwm -pi) pressure difference across openings (pwm is the external pressure due to wind and pi is the internal pressure) (Pa)
r = air density (kg/m3)
The ventilation rate (n), which is defined as the volumetric air flow rate (Q) entering a space divided by the space volume (v), is normally determines the number of times per hour that the space air is replaced by external fresh air.
Many researchers have carried out qualitative studies of the pattern of indoor air flow caused by wind pressure. However, quantitative evaluations of indoor air flow especially in tropical conditions are still rare. Therefore, this paper is intended to present new findings in this area of specialisation. This paper covers only the first stage of a series of investigation i.e. evaluation of the effect of orientations, height of buildings, external facade treatment, special features, and various combinations of internal atria and wind scoops on the external pressure and flow patterns. The variations of the Cp values will give an earlier indication on the effectiveness of the wind-driven ventilation. Further research on the effect of porosity on the pressure values and the actual prediction of the internal air flow will be covered in different papers.n = Q/v (m3/h)/(m3) ---------------------(viii)
The required pressure measurements on the selected building were conducted in the ASR&D, UWCC atmospheric boundary layer (ABL) wind tunnel. This open circuit tunnel has an overall length of 11.61 m and a working section 2 m wide, 4 m long and 1 m high. In the first stage, which applied only to the isolated buildings, the individual models were placed in turn at the centre of a turntable with its aluminium base plate tightly screwed to the floor of the turntable. A hole was provided at the centre of the turn table’s floor which enabled all the silicon rubber tubing from the tappings to pass through. Each tube was then connected to the electrical manometer which was linked directly to the data logger. The reference pressures were obtained from the "pitot" static probe located at a height of 0.8 m from the tunnel’s floor. Both the surface pressures (p1) and the reference pressures (pref) were read simultaneously from the mean values of 480 readings in 30 seconds. Each set of p1 & pref readings consisted of 10 mean values of pressures, minimum and maximum values, and the standard deviations or "rms". The variations in the orientation (0o, 45o & 90o) were achieved by rotating the turntable to the desired direction. The fan speed was maintained at quite a high level (10 m/s) at the region of Reynolds number Re > 3000 throughout the whole experiment
The second stage of the experiments was concerned only with a few selected examples of the urban proximity models 1 & 2. The experiments were limited to a single wind direction (q = 0o), perpendicular to the main building facade. A similar instrumental and hardware set-up to that of the isolated buildings test was used in the urban proximity models test, and a similar scale factor of 1:200 was employed. . Appendix D shows external view of the ASR&D, UWCC wind tunnel and the typical set-up of the scale-models for the urban proximity models test in the wind tunnel.
The interaction between external wind and building geometries was analysed and presented in the form of pressure coefficient (Cp) values and flow patterns. The Cp values used in the analysis were derived from the wind tunnel pressure values of the "bluff body" models. Power law relationships with reference to the 1.5 m/s mean Malaysian meteorological wind speeds were used as one of the external design conditions. The actual pressure values derived from "bluff body" models did not include the porosity effect of openings, which was determined in separate set of experiments. The summary of the vertical distribution of Cp profiles of the scale-models are shown in Appendix E. Appendix F shows flow patterns
The findings of the test results may be summarised as follows:
Stage One: Isolated buildings.
1. Group 1 (100 m tall simple rectangular plan buildings consisting of Models 1b, 2b, and 3b):
 | The pressures exerted on the windward facade were always positive with the maximum values registered at 0.8 to 0.9 of the height ratio. The pressures on the leeward facades were always negative (suction). |
| Wind impinging normally to the long facades of the flat-faced building, the building with 1.5 m simple horizontal shading devices, and the building with 1.5 m vertical and horizontal shading devices resulted in almost the same distributions of pressures at 0o wind direction. These were indicated by the Cp values registered at the windward and leeward facades of the buildings. |
| Wind impinging obliquely to the long facade resulted in higher pressures being exerted at the down wind end of the building with shading projections than in the case of the flat-faced building. The corner projections of the solar shading devices caused stagnant positive pressure to be accumulated at the down wind ends of the main facade. |
| The negative pressure values exerted on the leeward facade of the three buildings were found to be almost identical for the normal and oblique winds. |
| For the elongated rectangular plan building, the wind-induced pressure generated a sufficiently large pressure difference across the windward and leeward facades. For normal and oblique winds, the differences in pressure were found to be capable of generating a good cross-flow of air. |
| Wind impinging parallel to the long facade resulted in almost equal negative pressure (suction) being exerted on both the opposite long facades of the buildings. The patterns of the pressure distribution were almost identical for the three buildings except that the flat-faced building had lower suction at the down wind end. Fluctuations in the pressure values exerted on both the long facades generated pressure difference, but this was of little significance in inducing a cross-flow of air. |
2. Group 2 (50 m tall simple rectangular plan buildings consisting of Models 1a, 2a, and 3a):
The vertical distributions of Cp values on the 50 m tall buildings that had similar facade treatments as the 100 m tall buildings were found to be almost identical. However, the pressure values of the lower buildings were smaller. These results revealed that the gradient velocity of the boundary layer wind had a quarter of the effect on the taller buildings when compared with the lower ones.
3. Group 3 (100 m tall buildings with special features and consisting of Models 5, and 6):
a) Model 5 - 100 m tall building with open central corridors.
| The pressure exerted on the windward facades of the front block were positive, whereas the pressures exerted on the leeward and other facades were negative except for the case when winds were parallel to the long facades. |
| Building with central corridors and open courts (such as Model 5) generated higher Cp values at the facades along the corridors for 90o wind direction. The pressure differences across the opposing facades at the middle level indicated that a better cross-flow could be generated across the building for winds blowing parallel to the long facades. |
| The distribution of pressure coefficients and pressure differences generated from normal and oblique winds indicated that a good cross-flow of wind could be generated across the front block, with a weak cross-flow across the rear block. This was shown by the large positive pressure differences across the front block and smaller negative differences across the rear block. |
b) Model 6 - 100 m tall building with external atria.
The atria consisted of four segmental recesses and were placed at the four corners of the building.
| The distributions of pressure coefficients on each facade and the flow patterns were found to be unique and were determined mainly by the building geometries and orientations. |
| When wind impinged normally to the main facade, the big segmental recess located at the top end corner of the windward facade was able to generate a significant reduction in pressures; whereas the atrium placed on the leeward side was not able to affect the pressure distribution of the whole facade. When it was placed at the side facade, this big external atrium was found to be capable of replacing the suction with positive pressure. |
| When oblique wind impinged directly into the atrium placed at the windward corner of the building, the atrium was found to be capable of forming stagnant positive pressures. The atrium acted like a wind scoop by collecting winds which could then be diverted into the building for natural ventilation if openings were provided. |
| With oblique winds, atria located at the down wind corners were able to increase the suction effect and displaced low positive pressures on the facade with suction. |
4. Group 4 (50 m and 100 m tall buildings with central atrium Model 4a and 4b):
| The pressures generated from normal and oblique winds were found to be positive on the windward facade and negative on the other facades of both buildings. |
| The result could be inferred that a narrow tall atrium located in the centre of a tall building together with a through opening at the ground level generated a suction effect to the whole atrium. In terms of ventilation strategy, this phenomenon could induce a stronger cross-flow to the front block, but a lower cross-flow across the rear block. This was clearly indicated by the higher positive pressure differences across the front block and lower negative differences across the rear block. |
| Oblique winds were able to be deflected inwards into the top section and into the bottom of the atrium. This effect was caused by the greater gap across the recess of the atrium formed by the diagonal dimension of the atrium. |
5. Group 5 - 50 m tall buildings with atrium and alternative wind scoops {Model 4a + (W/S 1, 2a, 2b, 3a, 3b, 4a and 4b)}
| Comparative evaluations were carried out among the seven combinations in order to select the most promising combinations. The combinations of Model 4a + W/S 2a and Model 4a + W/S 4a were shown to be capable of generating a consistent positive pressure difference throughout the whole height. These were essential in generating a consistent cross-flow of air across the building and throughout the whole height. The combinations represented the deep plan 50 m tall atrium building with 30o and 45o double inverted pitched roof wind scoops. |
| Other combinations were suggested as being useful in other conditions such as in weak wind conditions or in mixed environmentally controlled buildings. |
6. Group 6 - 100 m tall buildings with atrium and alternative wind scoops {Model 4a + (W/S 1, 2a, 2b, 3a, 3b, 4a & 4b)}
Similar comparative evaluations were carried out on the seven combinations of the 100 m tall building and the alternative wind scoops.
| The combination of a 100 m building with internal atrium and a simple flat roof wind scoop exhibited a similar performance to the building without a wind scoop. In both situations the atrium retained its suction effect which was capable of inducing an outflow of winds from the atrium. |
| By making the top section of the wind scoop slope downwards, the atrium became completely pressurised. |
| Installing a vertical divider in the middle of the wind scoop, thus blocking the direct cross-flow, caused the winds to deflect downwards and exert more pressure at the base the atrium. |
| The most promising combination which produced a good positive pressure difference across the whole front and rear blocks was the combination of Model 4b + W/S 2a. |
Stage Two: Buildings in Urban Proximity Models 1 and 2 (UPM1 and UPM2).
1. Group 1 (Model 1b in UPM1 and 2):
Urban Proximity Model 1 (UPM1) was intended to represent a high-density built-up urban area. The structure of a 100m tall building was surrounded by closely packed buildings, mostly of 0.5 of the height of the measured building and laid in staggered plan arrangement. Urban Proximity Model 2 (UPM2) consisted of a similar tall structure surrounded by a mixture of tall and medium-rise buildings of 50 m and 100 m. In the layout, a similar tall structure was intentionally located immediately opposite on the windward side in order to give a complete sheltering effect.
| The lower surrounding buildings enclosing a simple rectangular tall building caused very little effect on the pressure distributions of the top half of the windward facade of the measured building. However, the sheltering effect caused the pressure at the bottom half to be reduced tremendously except at the lowest levels closest to the ground. |
| The flow patterns surrounding the tall building showed that the sheltering effect caused by the low surrounding structures affected the pressures on the lower section up to almost half of the height. However, this sheltering effect had little significant effect on the protruding section of the tall building. |
| These results implied that a tall building surrounded by lower surrounding buildings could retained a high pressure difference of wind across the whole height except the lower section of the building. |
2. Group 2 ( Model 4b, Model 4b + W/S 2a & Model 4b + W/S 2b in UPM1):
a. Model 4b
| The pressure on the windward and leeward facades of the 100 m tall building with central atrium was affected significantly by the surrounding tall buildings. The top half of the front block could still have good cross-flow of wind, but the rest of the building suffered from low pressure differences that could not generate a good cross-flow of air. The entire atrium was found to be dominated by suction. |
b. Model 4b + W/S 2a and Model 4b + W/S 2b
| The building with an atrium and a double inverted pitched roof wind scoop caused the atrium to be pressurised. This was shown by the results for the two models (Model 4b + W/S 2a and Model 4b + W/S 2b). |
| The middle level of the front block was found to suffer from very low pressure differences that could lead to difficulty in generating a cross-flow of air. |
| The pressure differences across the rear block were high and capable of generating a good cross-flow of air in the same direction as the prevailing wind. |
3. Group 3 ( Model 4b, Model 4b + W/S 2a & Model 4b + W/S 2b in UPM2):
| Surrounding buildings of equal height, especially those placed in close proximity and on the windward side, severely affected the pressure values of the building. The atrium of the building was found to register almost zero pressure values. |
| Placing of wind scoops over the roof did not much improve the situation when very small pressure differences were generated throughout the whole buildings. |
| The sheltering effect of the nearby windward buildings resulted in the skimming flow of wind around the buildings. |
| This phenomenon could be one the major problems in an extremely dense town centre where closely packed tall structures of almost equal height limit the use of natural ventilation. |
From the wind tunnel experiments on "bluff body" models it can be concluded that:
| The orientations and external geometries of high-rise buildings determine the wind-induced pressures over the building envelopes. |
| For cases of isolated tall buildings, greater building height produces increased exposure to wind and therefore a higher wind-induced pressure. |
| For normal winds the pressure on the windward facade reaches a maximum at about 0.8 to 0.9 of the height ratio, with mean Cp @ 0.9 ± 7%, and a minimum at about 0.1 of the height ratio, with Cp @ 0.5 ± 0.7%. The pressures on the leeward facade are negative (suction) with mean Cp @ -0.4 ± 8%. |
| For oblique winds the pressure on the leeward facade reaches a maximum at about 0.7 to 0.8 of the height ratio with mean Cp varying form 0.2 to 0.9 ± 13%. |
| For parallel winds the pressure on the main facades is negative with mean Cp @ -0.5 ± 30%. |
| Simple shading projections do not influence the pressure values for normal wind, but give higher pressure values (+ 30% to 60%) at the down wind corner for oblique winds. |
| High-rise buildings with internal atria, through openings at ground level and inverted double pitched roof wind scoops are capable of creating positive pressure differences across the front and rear blocks of a deep plan building. |
| High-rise buildings in closely packed urban surroundings consisting of lower buildings are capable of retaining good pressure differences across most of the building except at the lower part, where pressure is reduced by @ 60%. |
| High-rise buildings in closely packed urban surroundings consisting of similar or taller buildings are less capable of retaining sufficiently high pressure differences for a natural cross-flow of air. Generally, at 0.8 height ratio, the pressure is reduced by about 60% to 70% compared to the pressure on an isolated building. |
| Thus, for buildings in an open built-up area, natural ventilation is possible but becomes more difficult in a highly built-up area. |