Airflow around buildings consists of natural winds that travel around and possibly through buildings. Airflow around buildings has two influences on industrial ventilation:

1. Wind pressures exerted on the exterior building surfaces, which can influence air movement indoors

2. The outdoor movement of air contaminants, which can degrade indoor air quality if brought indoors with insufficient dilution.

This section will describe general features of airflow patterns and then present information on the dimensions and locations of recirculating (stag­nant) zones around the building envelope, which determine wind pressures and contaminant dilution. This knowledge allows one to select the locations of stacks and air intakes and to calculate infiltration and natural ventilation rates. General Features of Airflow around Buildings

Buildings are immersed in an atmospheric boundary layer in which the wind is influenced by friction with the earth’s surface. In this layer, wind speed tends to gradually increase with height and turbulence levels decrease with height, as described in many texts.1 Surrounding buildings, terrain, and vege­tation strongly influence wind and turbulence at a building site. The wind and turbulence levels are also influenced by thermal stratification of the atmo­sphere, such as ground-level inversion layers. The major parameters of the wind at a building site depend on the Reynolds (Re), Karman (Ka), and Rich­ardson (Ri) dimensionless characteristics:

Re=Vl/v Ka = J[V7)2/V Ri = gl/p (dp/dZ)/(d V/dZ)2, (7.227)

Where / = building characteristic dimension (height or width), v = kinematic viscosity, V’ = velocity fluctuation, V mean velocity, p = air density, dp/dZ = vertical density gradient, and dV/dZ = vertical velocity gradient.

Winds traveling past a building will be greatly modified compared with winds in the absence of the building. Hosker reviews the information on flow around both isolated structures and building clusters.2 Snyder and Lawson present detailed trajectories around isolated buildings obtained with detailed flow measurements.5

Figure 7.96 illustrates typical flow patterns for wind directly approach­ing a building face. Airflow in the undisturbed zone has a speed profile de­pendent on the terrain roughness and the level of atmospheric stratification. Obviously, most wind will be deflected around and over the building. Wind


* wind


FIGURE 7.97 Flow pattern around a long rectangular building (L > 2.5H). (Reproduced from Wilson 1982.)

Xc = 0.5 R Lr = 1.0/J,










R =


Where B$ = min (H, W], B, = max [H, W], and H and W = upwind building face height and width, respectively.

[f the building has significant length L in the windward direction (Fig. 7.97), the flow will reattach to the building and may generate two distinct re­circulating zones—on the building and in its wake. In the case of a long build­ing (L > 2.SH), the recirculation zone created by separation of the flow at the front edge of the roof extends to some distance, La smaller than the length of the building {Lc — 0.9R). Beyond this zone, at a distance of approximately 10H to 12H, the flow streamlines along the roof surface become similar to those absent the effect of the building.

For W/H < 10, the length of this recirculating zone is reduced. The val­ues of Lr and Hc can be calculated using the reduction coefficient Cr from Ta­ble 7.30.

Winds approaching a rectangular building at an angle will have different flow patterns than winds directly approaching a building face. Figure 7.98 il­lustrates the case for approach from a 45° angle. On the roof a pair of hori­zontal, counterrotating vortices emanate from the upwind roof corner. The negative (suction) surface pressures near the upwind corner can be intense, several times the magnitude of the dynamic pressure of the approaching wind.



Estimation of building pressure comprises two steps:

1. Determination of the reference wind speed for the building site, vm

2. Determination of the pressure coefficient, Cp, for the particular location on the building

Wind Speed

Wind has a highly turbulent and gusting character. In addition, a time- mean speed varies with the height from the ground and the roughness of the terrain over which the wind passes. The time-mean wind speed profile can be determined using the following expression:

Vm/v met = cH‘ (7.234)


V„ = mean wind speed at height H above the ground, m/s vmet = mean wind speed measured at a weather station, normally at a height of 10 m above the ground, m/s c — parameter relating wind speed to nature of the terrain (see Table 7.31) a = exponent relating wind speed to height above ground

To evaluate the wind speed at height H it is necessary to know the value of vr for the required location. This may be obtained either from a local weather sta­tion or from wind contour maps of the country. Normally, vr represents the hourly mean wind speed that is exceeded 50% of the time at a particular site.

Other equations describing the wind profile are available from ASHRAE7 and from Sherman and Grimsrud.8

Surface Pressure Coefficient

The second part of computing building pressures involves the pressure coeffi­cient for a particular spot on the building. The surface pressure coefficient, Cp, in­dicates the share of the wind kinetic energy that is transferred to the static pressure:

Cp = 2P/(pvm2). (7.235)

The value of coefficient Cp at the point on the building surface changes within a range of -2 <Cp*l and is determined by

1. The building geometry

2. The wind velocity (i. e., speed and direction) relative to the building

3. The location of the building relative to other buildings and the topography and roughness of the terrain in the wind direction

4. The location of a point on the building envelope

TABLE 7.31 Terrain factors for Eq. (7.234)




Open flat country



Coun try with scattered wind breaks









The surface pressure coefficient is normally derived from pressure measurements in wind tunnels using reduced-scale models of buildings or building components, or from pressure measurements in the actual build­ings. In general, the Cp values depend upon the Re, Ka, and Ri numbers. However, experimental tests are typically conducted at high values of Re (Re s 1000), in isothermal conditions (Ri = 0), and considering self-simi­larity against the Ka number (Ka = idem). Most test conditions allow only geometrical scaling.

For a building with sharp corners, Cp is almost independent of the wind speed (i. e., Reynolds number) because the flow separation points normally oc­cur at the sharp edges. This may not be the case for round buildings, where the position of the separation point can be affected by the wind speed. For the most common case of the building with a rectangular shape, Cp values are normally between 0.6 and 0.8 for the upwind wall, and for the leeward wall

0. 6 < < —0.4. Figure 7.99 and Table 7.32 show an example of the distri­

Bution of surface pressure coefficient values on the typical industrial building envelope.

Values of Cp for simple building geometries may be obtained from the Brit­ish Standards Institution6’9 or from Liddament.10 The following relationship between wind incident angle a, building side ratio, and average surface pres­sure coefficient is based on the database developed by Swami and Chandra:11

NCp = ln[1.248 -0.703sin(ot/2) — 1.175sin2(a) + 0.131sin3(2aG)

+ 0.769cos(a/2) + 0.07G2sin2(a/2) + 0.717cos2(a/2)],



NCp = normalized Cp

A = angle in degrees between wind direction and outward normal of wall under consideration

G = natural log of ratio of width of wall under consideration to width of adjacent wall


TABLE 7.32 Approximate Surface Pressure Coefficient Values for a Building with a Rooftop Vent



Cp | and Cpj





> 2













+ 0.3





+ 0.8

+ 0.8

+ 0.8

S 60









< 0.5


> 2










A detailed method of determining pressure coefficients is to perform ex­periments with a wind tunnel facility. Cochran and Cermak compared wind tunnel pressure coefficient measurements with field measures on a test build­ing and found excellent results, with the exception of small areas beneath the vortices near the upwind roof corner for winds approaching at 45°.12 For infil­tration and natural ventilation designs, wind tunnel results should be suffi­ciently accurate.

Another detailed method of determining pressures is computational fluid dynamics (CFD), which uses a numerical solution of simplified equations of motion over a dense grid of points around the building. Murakami et al.13 and Zhoy and Stathopoulos14 found less agreement with computational fluid dy­namics methods using the k-e turbulence model typically used in current com­mercial codes. More advanced turbulence models such as large eddy simulation were more successful but much more costly.13

For simplified buildings or in cases where detailed modeling is not practi­cal, simplified tables of coefficients are presented by Liddament for low-rise buildings with two building shapes, for open and sheltered buildings, and for various walls and approach wind angles.15 ASHRAE also summarizes results from other studies.7 Contaminant Transport around Buildings

Transport of outdoor contaminants is controlled by both the mean mo­tion of winds and dispersal by turbulence. Since airflow around buildings has distorted wind trajectories and enhanced turbulence compared with the air­flow without buildings, contaminant transport requires special consideration in the presence of buildings.

Contaminants can either be generated by sources within the building or be emitted from exhaust stacks or nearby locations. The goal is to maintain indoor air quality by minimizing the entry of outdoor contaminants into the interior of the building through careful design and location of exhausts, in­takes, and building openings. Since most industrial exhausts and air-handling systems are located on the roof, the airflow on the roof is of great interest. As discussed earlier, when wind directly approaches a building side, a separation zone is created on the roof near the upwind edge. This zone will have high turbulence intensities and greatly distorted streamlines. In fact, flow reversal occurs, in which the airflow is in the opposite direction to the approaching wind. Any building exhaust emitted into this zone is quickly dispersed and spread over the roof surface, creating large rooftop contaminant concentra­tions. It is especially important to avoid placing low-momentum exhausts and air intakes within the same recirculating zone on the roof or side of the building.

Furthermore, winds at an angle to the building side will have downward air motions above the roof and downwind of the building, which also will in­crease rooftop concentrations.

Therefore, to avoid separation and high-turbulence zones on the roof, contaminant exhaust systems should be designed to extend as high above the roof as practically possible. Two ways to achieve high exhaust trajectories are with the use of taller stacks and greater exhaust vertical momentum, which increases the throw or rise of the exhaust. Tall stacks are an obvious solution but have aesthetic problems, even in an industrial setting. Achieving greater vertical momentum can be gained in several ways:

• Avoiding raincaps over exhausts, which eliminate vertical momentum

• Increasing exit velocity by narrowing the exhaust opening

• Manifolding separate exhausts into fewer stacks

• Placing exhausts very close together so that the exhaust plumes can merge (which can help exhausts that cannot be manifolded)

In the design of an industrial facility, a detailed analysis of contaminant transport can be beneficial in specifying exhaust designs and intake locations. As with building pressures discussed above, analysis techniques range from simplified models to experimental methods and sophisticated computational methods. The greater the building geometry complexity, the less useful are the simplified models.

One simplified method for determining stack height is a geometric method described in ASHRAE.7 The geometric method assumes an exhaust plume shape with a lower boundary having a 1:5 slope relative to the horizon­tal. The stack and plume are raised until the lower plume boundary is above rooftop penthouses, separation zones, and zones of high turbulence. ASHRAE provides equations for the sizes and locations of the separation and turbulence zones.7 A stack height reduction credit is provided to account for the vertical exhaust momentum.

Another method is a series of exhaust dilution equations based on Wilson and Lamb16 and a series of earlier papers summarized in ASHRAE.; This method is based on wind tunnel tests on simplified buildings and is intended to provide conservative (low dilution) results. Wilson and Lamb compared the model to actual field data collected at a university campus and found that the model did indeed predict dilutions similar to measured worst-case dilutions suitable for a screening model. However, many cases resulted in conservative underpredictions of dilutions.16

An alternative simple model for contaminant dilution of rooftop exhaust stacks is presented in Halitsky.17 This model combines a jet region specification for the upward exhaust movement with a more traditional Gaussian plume region controlled by atmospheric and building-generated turbulent dilution.

For computing dilutions in the downstream cavity wake (intakes or build­ing openings on the side of a building) due to contaminants from a rooftop stack, another model is presented by Schulman and Scire18 and is incorporated in a U. S. EPA screening model.19 More sophisticated analysis may be achieved with wind tunnel models. Saathoff et al. compared wind tunnel dilution pre­dictions with measurements on an actual building and found agreement within a factor of 2, a reasonable limit for dilutions, which can vary over many orders of magnitude.20 Higson et al. found greater differences between wind tunnel measurements and a miniature building in field conditions, corre­sponding to a factor of 4 or more.21

Computational fluid dynamics (CFD) is becoming more popular, as dis­cussed above for building pressures. However, a recent paper found difficul­ties in the practical use of current commercial codes due to the wide range of user inputs and decisions.22 Other papers are exploring alternatives to the standard k — e model typically used in commercial codes today.23,24