Introduction

The visualisation of complex flow patterns has played and always will play a very important part in developing an understanding of the flows that occur inside turbomachines. The visualisation may provide graphic details of a phenomenon such as the shock-wave pattern in a transonic turbine cascade or it may provide quantitative data in the form of interference fringes that represent lines of constant refractive index and so isoclines in the same flow. The purpose of this section is to present the more popular methods of flow visualisation in use at the Whittle.

It is important to remember that the purpose of most simple flow visualisation experiments is to aid the understanding of the physical processes involved. When used in this way, a poorly conducted experiment or an incorrect interpretation or misrepresentation may have disastrous consequences.

Oil and Dye Flow Surface Flow Visualisation

The oil-and-dye technique appears in many forms but each is a development of the same basic technique. This involves the mixing together of a selected pigment and oil, coating the surface to be studied, allowing the gas flow to create a pattern and then photographing the results. The wall shear stress acting on the oil-and-dye mixture, that covers the actual surface, causes the mixture to move in the direction of the wall shear stress vector. As such, the oil-and-dye technique produces aerodynamic data The general objective of oil-and-dye flow visualisation is to prepare a mixture of such consistency that under the test conditions, it form a clearly defined pattern of streaks that can be photographed and yet that are also fine enough to produce the required detail.

The patterns develop because the pigment is deposited on the surface as the mixture flows by, in what is in effect a silting process behind small concentrations of pigment in the mixture or depositions on the surface. The mixture flows around these concentrations and some of the pigment is deposited in the wake. The deposits continue to grow and so long streaks develop. Eventually all the oil will be blown from the surface and the pattern is left in a relatively dry state that can then be recorded. The size of the pigment particles and their dispersion within the oil govern the detail of the final patterns.

At the Whittle Laboratory, two versions of the technique are in frequent use. One is used in the continuously running transonic test facilities while the other has been developed for use in low speed large scale cascades and rotating rigs. Both techniques use the same pigment. This is fluorescent paint pigment with a particle size of about 1 mm. For use at high speed, the pigment is mixed with a silicone oil; for low speed use, light diesel oil is employed. No additives are employed. These pigments give a high-contrast pattern irrespective of the background colour and provide very high quality photographs without difficulty. Typically, the patterns in both facilities take about ten minutes to develop. The patterns are usually illuminated with Ultra-violet light. The rendering of the surfaces non-reflective with a non-reflective finish such as matt-black cellulose is particularly important when the surfaces are metallic.

Figure 1 shows a perspective photograph of the suction surface and the film-cooled end-wall of a turbine cascade. The upstream boundary layer separates and rolls up to form what is often called the horse-shoe vortex. This separation identified as the horseshoe and passage vortices. This is an example of the three-dimensional separation of an inlet boundary layer. When the flow is viewed looking along the separation line, then one obtains a two dimensional image similar to that at a two-dimensional separation point; the three-dimensionality arises because there is movement parallel to the separation line. The flow visualisation also shows the interaction of the cooling flow with the endwall boundary layer flow.

It is important to remember that while these patterns are real, they merely represent that flow that lies very close to the surface. The near surface flow invariably contains very low momentum fluid so that it responds very quickly to pressure gradients, be they normal or parallel to the mainstream direction. Thus, separation lines form at the surface in diffusing flows while overturning takes place near the end-wall of a turbine cascade. Much of the flow in the above mentioned turbine cascade does not therefore behave like that close to the end-wall but instead passes through as if it were in a two-dimensional cascade without end-walls. Indeed, even the end-wall boundary layers do not really behave as the flow visualisation would suggest, being instead skewed so that there is a gradual transition from the free-stream to the surface. Surface flow visualisation of this type may therefore be regarded as representing the flow in the worst possible light.

Mixutures of coloured paints can be used to highlight different features of the surface flow patterns. Figure 2 shows one such example.

Schlieren Photography

Optical methods have one distinct advantage over any other technique; they are non-intrusive. A further advantage is that the upper frequency limit of the system is essentially governed by the speed of the photographic film or electronic camera so that extremely rapid flow variations can be studied with ease.

Only optical methods that are used to study the density fields in transparent fluid will be described below. All rely upon the fact that the speed of light or the refractive index is dependent on the density of the fluid. In practice the approximate relationship

r = (n-1)/k; k = constant

where r is the fluid density and n the refractive index hold true for most gases of practical importance in turbomachinery. The constant in the equation is usually given by reference conditions r₀  and n₀.

Since the refractive index varies with the wavelength of the light, monochromatic sources are usually employed. These are usually either the mercury line (546 nm) or the visible line from a Helium-Neon laser (633 nm). If the density varies within a fluid, then wave fronts of light are turned so that there is refraction and this is the basis of the Schlieren and shadowgraph methods; there is also a relative phase shift of the light rays and this is the principle of the interferometer. Shadowgraph systems are used to indicate the variations in the second gradient of the density field in the direction normal to the light beam; Schlieren indicates the gradients while interferometry responds to differences in the optical path length and so provides lines of constant density. All three techniques produce an image that represents the average history of the beam as it passes through the fluid. In effect, the image represents the integral along the path length of the quantity under examination. As such, these systems are best suited to two-dimensional investigations. Each system has its own level of sensitivity.

The question arises as to what flow features give rise to changes in the fluid density. In turbomachines, the flow is often compressible so that there are always density variations within the flow. Interferometry will measure these changes directly. Schlieren and shadowgraph systems require more rapid changes etc. but even the poorest of systems will not fail to indicate the location of shock waves.

Howard Hodson

Home | Current Research | Research Opportunities

Publications | Staff and Students

Research Facilities | Travel Information | Contact Information | Links