Bluff Body Aerodynamics: Features and Development

In this article we will discuss about the primary features and historical development of bluff body aerodynamics.

Primary Features of Bluff Body Aerodynamics:

The term “bluff body aerodynamics” is intended to cover the study of flow fields around, and force on, a body from which boundary layers are separating to form unsteady vortex flows in a wake region. The great majority of body shapes in the field of wind engineering involve bluff body aerodynamics as distinct from the attached flow aerodynamics associated with streamlined aeronautical shapes.

The study of wind engineering phenomena has as its basis the study of bluff body aerodynamics. In a paper “Turbulence and the Leading Edge Phenomenon”, Melbourne, 1992, discussed the behaviour of shear layers separating from the leading edge of a bluff body and the influencing role played by free-stream turbulence, in particular the effect on the pressures developed under the reattaching shear layer. This review will lean heavily on this earlier work to describe the origins of bluff body aerodynamics to provide support for the wind engineering applications presented in this volume.

The significant characteristics of the flow field around a bluff body can be described with reference to Fig. 1 showing the time averaged pattern of flow past a square cylinder. The upstream flow divides around a stagnation streamline and then separates either at some sharp edge or where the pressure gradient becomes sufficiently adverse.

Upstream and outside the separated shear layers the flow is essentially irrotational whilst under the shear layers an unsteady, low velocity wake region is formed. Reynolds number, and surface shape, and roughness, along with free-stream turbulence, also play a significant part in determining how the shear layer behaves.

The aerodynamic forces on a bluff body in a turbulent flow can be conveniently divided into those which cause along-wind response and those that cause cross-wind response.

From the work primarily of Davenport (1967) and Vickery (1971) it can be concluded that the along-wind response of most structures originates almost entirely from the action of the incident turbulence of the longitudinal component of the wind flow (superimposed on a mean displacement due to the mean drag).

Their analytical methods, using spectral and spatial correlation considerations to predict the along-wind response of many structures has become highly developed, to the point where the gust factor approach is included in a number of wind loading codes. By comparison, the mechanisms causing the cross-wind response have proved to be so complex that as yet there is no simple generalised analytical method available to calculate the cross-wind response of structures.

Historical Development of Bluff Body Aerodynamics:

The first known documented appreciations of the effects of turbulence on the pressure on a stream-wise face under a separated shear layer had their origins in wind engineering studies, as they came from a comparison of full scale and model scale pressure measurements on small buildings.

Bailey (1933), working at the National Physical Laboratory in England measured wind pressures on a railway shed in full scale and compared them with pressures measured on a wind tunnel model of the shed in relatively low turbulence uniform flow. The measured pressure coefficients were not in good agreement, particularly near the windward eaves.

The difference can be attributed to the difference in the approach roughness and consequent turbulence of the incident flow. Bailey et al (1943) undertook some more wind tunnel measurements on a model immersed in the boundary layer on the floor of a large wind tunnel and there was much better agreement between the model and full scale in the boundary layer flow than in the low turbulence uniform flow.

In particular they noted the effect of the ratio of the height of the building, h, and the boundary layer roughness length, Zo, which Jensen (1958) later proposed as a significant modelling scaling ratio. Jensen demonstrated the dependence of the pressure distributions around a bluff body on the boundary layer, or more specifically as we know now on the turbulence parameters, in a very similar experiment to that of Bailey in that he measured pressure distributions on a shed in full scale and in four scaled boundary layers.

The effect of increasing the turbulence level at eaves height, as the model boundary layer roughness length increased, was shown to move the peak low pressure forward towards the leading edge. For low turbulence the flow separates and does not reattach such that the low pressure over the roof (and leeward wall) is nearly constant in the recirculating wake region. For high turbulence the flow separates at the leading edge and reattaches on the roof and in this region very low pressures were seen to occur (as they did in full scale) with a significant rise in pressure beyond reattachment.

In the 1960’s the work of Jensen and Davenport led to the development of boundary layer wind tunnels in which models of buildings and structures were tested. Using Jensen’s Law meant that the model measurements of pressure and force on bluff bodies were being conducted in properly scaled turbulent flow, and hence the effects of turbulence on these parameters were being reasonably well covered by those using these modelling techniques. There were several studies in this period which did lay the foundations for much of the more recent research.

The first of these was a study by Vickery (1966) of the effects of free-stream turbulence on the flow around a square cylinder. Whilst this study was aimed primarily at the fluctuating forces (and it was found that the addition of turbulence greatly reduced the fluctuating lift) it was shown that the incident turbulence had a significant effect on the mean and fluctuating pressure fields and that the span-wise correlation of pressures was also significantly lower in turbulent flow.

Bearman (1972) showed that the base pressure behind square and circular flat plates decreased with increasing turbulence (and integral scale of turbulence) to present the paradox with Vickery’s measurements of base pressure behind a square cylinder which increased with increasing turbulence.

It was Gartshore (1973) who provided the first incisive description of the effects of turbulence of flow around bluff bodies and who provided an explanation of the above paradox. The work of Gartshore is worth dwelling upon, as it was not widely published at the time, and to this day the critical part of the effect he demonstrated to be due to the fine scale turbulence is not well understood.

Gartshore (1973) suggested that the effect of increasing free-stream turbulence is to increase the turbulent mixing in the shear layers and hence to increase the rate of entrainment from the wake and decrease the radius of curvature of the shear layer. Gartshore further showed that fine scale turbulence, produced by a thin rod on the stagnation streamline, is sufficient to cause these effects.

Gartshore’s graphical descriptions of the paradoxical effects of free-stream turbulence are given in Fig. 2 and his conclusions are as follows:

“Small scale free stream turbulence can influence the drag, fluctuating lift, base pressure coefficient and galloping behaviour of bluff bodies by increasing the rate of entrainment of the shear layers shed from them. The free-stream turbulence, after severe distortion by the mean flow field, produces the increased mixing and entrainment by its presence in the shed shear layers, probably in the form of local concentrations of stream-wise vorticity.

Only turbulence approaching a bluff body along its front stagnation line is required to produce the major effects of free-stream turbulence on the flow around the body, so that a narrow wake effectively replaces grid-produced turbulence for experimental purposes. This observation also emphasises the importance of nearby upstream structures when considering the flow around tall buildings in the natural wind.

Turbulence can produce increased drag and unstable galloping oscillations in shallow rectangular bodies (b/d < 0.5) which are stable in the absence of turbulence.

It can also decrease the drag and stabilise bodies of deeper section b/d > 0.6 which are unstable without turbulence. These effects are explicable qualitatively in terms of the increased entrainment resulting from turbulence.

The effect of turbulence scale is small provided this scale is significantly smaller than the body dimension.

Fig. 2 Sketches of Expected Effect of Turbulence on Dividing Streamlines as it Affects Drag and Lift of Rectangular Sections.

The key observation of Gartshore was the recognition that increasing turbulence caused the radius of curvature of the separated shear layer to reduce. This has the most important effect of causing earlier re­attachment of the flow on a stream-wise surface, if it is present, which leads to lower pressures on the stream-wise surface near the leading edge and higher base pressures (and lower drag) and can change lift curve slopes from negative to positive, thus impacting on the galloping instability phenomenon. Virtually all the stream-wise surface and cross- wind effects of turbulence on bluff body aerodynamics can be attributed directly or indirectly to this process.

Lee (1975) extended Vickery’s pressure measurements on a square cylinder in increasing free-stream turbulence and in particular noted that increasing turbulence caused the shear layers to thicken and reattach intermittently onto the stream-wise surfaces in support of Gartshore’s conclusion.

He also noted that the axial correlation of the stream-wise surface pressures was relatively unaffected by increasing turbulence intensity up to σu√u = 0.08, but that there was then a significant reduction in correlation as the turbulence intensity increased to 0.12, (presumably because the highly three dimensional intermittent reattachment process was becoming established). Laneville (1975) in flow visualisation experiments also showed that an increase in turbulence intensity caused earlier reattachment of the separated shear layers on rectangular prisms.

Melbourne (1975) after several years of observing the occurrence of very low intermittent, peak pressures on stream-wise surfaces near the leading edge of tall building models, and influenced by the conclusions of Gartshore (1973), advanced an hypothesis suggesting that the very low pressures which occurred under the reattaching shear layer under conditions of high free-stream turbulence were due to an instability process under certain conditions.

“The shear layers and pressure underneath fluctuate. For certain conditions of free-stream turbulence, angle of attack, (and radius of leading edge curvature) the shear layer commences to occasionally reattach onto the wall. As this occurs the cavity region under the shear layer is no longer vented and any decrease in pressure causes the initial radius of curvature of the shear layer to decrease further which in turn increases the local free-stream velocity outside the shear layer and further decreases the pressure. This is an unstable process which proceeds until the shear layer breaks up into a complete separation again and the cavity becomes vented. During this unstable process, very low, intermittent pressures can occur on the surface under the shear layer near the leading edge.”

This hypothesis was largely based on what could be seen with crude flow visualisation at the time and whilst the description of the dynamic behaviour of the shear layer with the occurrence of the very low pressures was correct the instability idea was shown by Saathoff to be incorrect and that it was the presence of strong vortices convecting downstream under the shear layer which caused the very low pressures and the appearance of early reattachment at the same time.

Melbourne (1979) further investigated Gartshore’s conclusions with respect to the importance of the fine scale turbulence, and in so doing developed a Small Scale Spectral Density Parameter to quantify the spectral density of the incident free-stream flow at scales the size of the shear layer width.

In particular he showed that increase of the small scale turbulence within flows with the same large scale turbulence, significantly increased the magnitude of the very low pressures under reattaching shear layers. He also showed that venting the bubble under the reattaching shear layer with a leading edge slot significantly reduced the magnitude of the very low pressures which had applications in reducing cladding wind loads on roofs and the response of cantilevered roofs.