Evaluation of Wind Loads on Low Buildings | Wind Engineering

In this article we will discuss about:- 1. Introduction to Wind Loads on Low Buildings 2. Historical Development of Wind Loads on Low Buildings 3. Codes and Standards/Full-Scale Measurements 4. Computational Evaluation of Wind Pressures.

Introduction to Wind Loads on Low Buildings:

Wind loads on low buildings have not received the attention they deserve when the large investment in such structures is considered. Unfortunately, they have an inconvenient way of reminding us of this neglect when a hurricane or tornado strikes.

Much of the data from which present wind standards and codes of practice have been derived were obtained from model tests carried out before all the implications and the importance of correctly simulating the atmospheric boundary layer were appreciated. Only mean pressures were measured and the dominating influence of fluctuating pressure was missed.

Pressure coefficients have been used with a variety of reference wind pressures associated with wind speeds averaged over several minutes or fastest mile wind speeds or gust speeds averaged over a few seconds due to the lack of understanding of the correct relationship of the pressures to the oncoming wind flow.

More recent studies have provided more useful indications of the important features of wind loading on low buildings. For example, it has been shown that the dominant loads are fluctuating and not necessarily organized either spatially over the structure or in time; effective loads are reduced with increasing tributary area but in different magnitudes depending on the location of the area vis-a-vis the wind flow; the surrounding terrain roughness and the adjacent buildings have an important influence on the distribution of pressures and the amplitude of the fluctuating components; and building shapes, in particular roof pitch and configuration, as well as architectural details (eaves, parapets etc.) are the primary factors influencing pressures.

The ongoing development of computational wind engineering is also promising for the evaluation of wind loads on low buildings. Numerical solutions have the potential to overcome several restrictions of physical simulation such as the Reynolds number limitations in accurate modelling flows over curved surfaces, the inability to fully model swirling flow impacts on structures (tornadoes) and some types of gust fronts (downbursts).

However, the ability of computational approaches to evaluate wind loads on low buildings has yet to be demonstrated for cases other than mean pressures for normal wind directions and simple building geometries.

Historical Development of Wind Loads on Low Buildings:

Much of the early research into wind loading was concerned with low buildings. Amongst the earliest wind tunnel tests performed were the experiments by Irminger in the 1890’s on small models of gabled houses. Between the wars most western countries undertook systematic studies of the aerodynamic pressure coefficients on different building shapes in wind tunnels.

Some of these studies were extremely thorough such as those by Irminger and Nokkentved (1930,1936) and those by Ackeret (1936) given in the Swiss Normen and presented in the 1965 edition of the National Building Code of Canada and other standards.

Unfortunately, the available wind tunnels and testing techniques at the time were those developed for aeronautical purposes. In these the flow was as smooth and uniform as possible; therefore it differed from the natural wind which is not only turbulent but also contains significant velocity variations with height, particularly along the height of low buildings.

The effects of such lack of similarity of the flow conditions in model and full scale were highlighted by a few full scale experiments such as those by Bailey (1933) on a railway car shed and those by Amstein and Klemperer (1936) on the Akron Airship dock. Both have indicated that the predictions of wind tunnel model tests, if they are carried out in uniform flow, can differ substantially from the full-scale realization.

Typical results from Bailey’s car shed are shown in Fig. 1. The data show the average and standard deviation of the pressure coefficients Cp for 24 occasions in which the wind was westerly and normal to the ridge of the roof. For this wind direction the wind approached the car shed over approximately 500 m of open field.

The diagram also shows the result of a wind tunnel test carried out in uniform flow without turbulence. Clearly in full scale the suctions recorded are generally significantly greater while the positive pressures are somewhat less than in the wind tunnel model. On the other hand, in case of Akron Airship dock, while positive pressures are similar, the peak negative suctions are significantly less in full scale than at model scale.

Although several researchers recognized the need for adequate simulation of the turbulence of the natural wind and its velocity variation with height, the requirement was not confirmed until Jensen (1958) proposed the “Model Law for Phenomena in Natural Wind”, i.e., scaling of the “roughness lengths” of the ground surface in the wind tunnel should be the same as that for the model itself.

To achieve this he roughened the floor of the wind tunnel using a surface of suitably rough texture – corrugated cardboard, sand paper, wood slats – depending on the full-scale surface being considered. The results of full-scale and model experiments carried out to demonstrate the effect of this scaling are shown in Fig. 2, in which the roughness length z_o characterizes the roughness of the surface.

It is particularly noticeable that the uniform flow case (h/z_o → ∞) produces radically different pressure distributions from the full-scale experiments but that the latter are in agreement with the model tests for the equivalent Jensen number h/z_o. An extension of Jensen’s model law is that the scales of turbulence, i.e. the characteristic eddy sizes, and the effective boundary layer thickness should also be scaled. Following Jensen’s pioneering work, the wind tunnel tests generally respected these modelling laws.

It is worth mentioning that, although wind pressures on buildings reported before 1958 may not be representative for the reasons explained, some of the early ideas hold well up to date. As early as 1884 Sir Benjamin Baker stated that the mean wind pressure on a large area must be less than that on a small area because “threads of the currents moving at the highest velocity will strike an obstruction successively rather than simultaneously”.

Having that in mind Stanton (1924) invented the wind pressure recorder which could be considered as a mechanical mean pressure transducer as shown in Fig. 3.

The apparatus consists of a series of chambers, each containing a flexible corrugated diaphragm, the displacement of which, due to the difference of air pressure on its sides, is followed by a guide-rod carrying a light pulley. A thin platinum wire is fixed at one end and passes over the system of the pulleys shown, the free end carrying a pen in contact with a drum actuated by clock work.

The motion of the pen integrates the separate motion of the diaphragms, so that a record of the average pressure differences in the chambers is obtained. It was Stanton’s idea that led to the utilization of manifolds and the development of the well-known pneumatic averaging technique used widely in the evaluation of area-averaged pressures on the building envelope.

A dramatic illustration of the tributary area effect is presented in Fig. 4 where individual instantaneous peak suction coefficients are plotted against wind direction for the four comer pressure taps indicated.

The true effective averaged suction coefficients (solid line) is also given together with the arithmetic mean of the non-simultaneously registered peaks (dashed line).

The data indicate that the lack of correlation over the area and the high pressure gradients occurring over these regions (particularly for the critical azimuths) lead to significant reductions in the effective pressure coefficients applicable to comer areas.

Codes and Standards/Full-Scale Measurements of Wind Loads on Low Buildings:

The material already presented and reviewed originates from studies in boundary layer wind tunnels and most of these findings will be included in the subsequent issues of wind codes and standards. In particular, wind loads on eaves of gabled- roof buildings as well as on roofs of various configurations (monoslope, sawtooth, multi-gabled, stepped) will be included in the 1995 issues of the Supplement to the National Building Code of Canada and the American Standard ASCE-7.

Other national and international wind standards will probably follow, as in the past. Nevertheless, there still exist limitations in the extent to which physical modelling in the wind tunnel is satisfactory and provides adequate results.

Full-scale measurement results form naturally the critical test of acceptance of wind tunnel data. In this respect, Hansen and Sorensen (1986), Richardson and Surry (1992), Okada and Young (1992), Milford et al (1992) have compared full- scale measurements with wind-tunnel results and found that mean and r.m.s pressure coefficients are in general satisfactory.

However, peak pressure coefficients obtained in wind tunnel tests were smaller than the full-scale measurements depending on the measurement location. This might be due to the difference in intensity of turbulence at full-scale or to the difference in frequency characteristics of pressure measurement systems in the wind-tunnel.

Although the state-of-the-art is still inconclusive in this respect, it appears that more full-scale tests will be required to shed light in a number of uncertainties still associated with the simulation of interaction between wind and low-building models in the boundary layer wind tunnel. Therefore, facilities such as the Texas Tech full-scale laboratory are extremely useful for further investigation of wind effects on low buildings.

Computational Evaluation of Wind Pressures on Low Buildings:

In spite of the dramatic progress in the area of computational wind engineering (CWE), there are still several obstacles for the numerical evaluation of wind pressures on buildings. Difficulties for the numerical prediction of wind effects on low buildings include, but are not limited to numerical accuracy, boundary conditions and refined turbulence models. The most important of these areas is turbulence modelling and Murakami (1992) has exposed the shortcomings of various turbulence models and discussed their relative abilities.

Most of the CWE applications in the area of wind effects on buildings have utilized the finite difference approach and more specifically the Control Volume Method with a two-equation (k-ε) turbulence transport model in its standard form or with modifications. Results have been presented in terms of mean flow conditions with associated mean pressures on buildings or mean velocities in the building environment.

Simple rectangular buildings in both plan view and cross section have been used in most cases, for wind perpendicular to the building face; in fact most of the CWE applications refer to this particular wind direction since this is the simplest flow condition by taking advantage of symmetry. Difficulties arise for different wind directions.

Stathopoulos and Zhou (1993) have applied the k-ε turbulence model for the numerical evaluation of mean pressure coefficients acting on an L-shaped – stepped roof – building as a particular case of a computational code applying to any building configuration consisting of two different rectangular buildings. Computed and measured data for different wind directions have been compared for several roof points.

Figure 16 compares the profile of wind-induced pressure coefficients along the centerline and the edge-line of the roof with the experimental data from Stathopoulos and Luchian (1990). The agreement between numerical results and experimental data is very good for the centerline, with some underestimation of suction generally shown by the computed data along the edge-line.

Figure 17 shows similar results for 60° azimuth. Regarding the centerline, the general agreement is good, with some underestimation of negative pressures by the numerical simulation mainly in the low-roof region. As far as the edge-line is concerned, the pressures are not predicted properly. In particular, the suctions on low-roof sections are significantly underestimated by the numerical simulation.

Likewise, on the higher roof, the computed data show a smooth change, while the experimental data show a high gradient. This suggests that the k-ε model and the two-dimensional boundary conditions may not be representative for this high vorticity region and such predictions break down completely.

Other numerical simulation models and approaches are currently under development and, if successful, they may provide a significant thrust to the current state-of-the-art. However, this may require some additional time.