1. [Abstract]
2. [Introduction]
3. [Experimental Apparatus]
4. [Image Processing]
5. [Results and Discussion]
6. [Conclusions]
7. [References]

ABSTRACT

This paper is concerned with the study of fluid flows through a cylinder head of an automotive engine. The study is performed by means of flow visualisation and velocity measurement, using an optical technique named Particle Image Velocimetry. The cylinder head is a transparent model, which can be constructed from transparent resin materials by vacuum casting and/or rapid prototyping methods. The results highlight the general features of the flow, such as laminar movement, vortices, stagnation and turbulence. Flow velocities between 0.2 and 5 m/s, with a maximum error of 5\% can be obtained.

INTRODUCTION

Optical techniques applied to fluid flows are used for interpreting and understanding flow phenomena to give a qualitative insight into the flow structure. Combined with quantitative measurements of specific flow parameters such as velocity, density, pressure and temperature, they give an accurate and complete picture of the flow. Experimental data obtained from optical techniques and quantitative measurements are also employed to asses and validate computational fluid dynamics codes.

Particle Image Velocimetry (PIV) is an optical method which consists of recording images of illuminated particles within the flow-field. The fluid velocity is calculated from the distance travelled by a particle in a known time period between two pulses of a light source. The schematic experimental apparatus is presented in Fig. 1. Unlike conventional techniques which use physical measurement probes introduced in the flow, the micron-sized particles used as flow markers in the PIV application described here, do not interfere with the flow.

Fig. 1. Experimental principle and general arrangement for PIV

The PIV technique provides a quantitative, instantaneous, whole-field visualisation and two dimensional description of the flow. This technique has a large range of applications, from slow flows modelled in a laboratory environment to transonic and supersonic flows produced in industrial wind tunnels and turbine engines. Various experiments involving liquid flows in enclosed tubes and internal passages have also been reported in the literature.

The technical details of the apparatus used for the present experiments are described in the following section.

EXPERIMENTAL APPARATUS

The experiment was designed by OEL and Envisage Group for APTT Rover laboratories, taking into account the existing components of a visualisation system. The old system was improved in order to provide high quality visualisation as well as accurate velocity measurement of internal flows in automotive engine models. The aim was to study the fluid flow within the cylinder head of an engine and the data obtained to be used for improvement of the coolant passage design. Nevertheless, the method can be applied to other engine components, such as oil galleries, engine intake and exhaust manifolds.

Transparent Model

In real engines, internal flows are difficult to measure using optical techniques, due to the lack of optical access. To overcome this problem the whole assembly has been fabricated from transparent epoxy resin. The transparent models are built by casting, a technique which encapsulates the passage cores into a transparent RTV resin. Thus, when the core is removed, the resin becomes a mould in which a low melting temperature alloy is poured and allowed to solidify. The metal core is then covered in transparent epoxy resin which is later removed by melting. The solid transparent model is completed with fittings and fixtures to provide a true representation of the real component. The time taken to complete such a model is eight to ten weeks. A picture of the transparent model of the cylinder head is presented in Fig. 2.

Fig. 2 Transparent model of the cylinder head for the Rover L-Series Diesel engine

Another technique currently in use is to build models directly from CAD data using a laser lithography process. This technique, known as rapid prototyping, has been developed by Rover to allow engine model testing to take place prior to manufacture. The testing currently in use on such models employs optical techniques for measurement of mechanical and thermal stress\cite{Redf95} and mechanical vibrations of the components. The method is also used to construct the cores used in the casting method described previously. Further studies of different resin materials necessary to build rapid prototyping models which are clear enough to allow flow visualisation are currently undertaken.

The novelty of the approach used in the current experiment refers to the use of a special fluid which has a refractive index matched to that of the material of the model. This minimises the optical aberrations which would otherwise be created by the shape of the assembly and allows a good calibration of the images obtained. The fluid is seeded with 10 micron diameter hollow glass spheres and circulated through the model's internal passages.

PIV Set-up

The particles are illuminated with a 1 mm thin light sheet, delivered from an Argon Ion laser through a combination of fibre optic and negative cylindrical lens. Since only a fraction of 1/10 of the maximum power of the laser (4 Watt) is employed in the final beam, other low power alternative light sources can be used (e.g. laser diodes). In order to change the operating mode of the laser from continuous-wave to pulsed, the beam passes through an Acousto-Optic Modulator (AOM) crystal. Thus, the user has the versatility of switching between the two operating modes, in other words from visualisation to PIV.

This device has the ability to separate the laser beam into lines of different frequencies which can be turned ``on'' and ``off'' independently with a short response time. The duration for which the beam is ``on'' gives the amount of light that falls onto the particle and hence the brightness of its image. The duration for which the beam is ``off'' gives the separation between pulses. Both these time steps can be externally set from a signal generator/driver circuit which triggers the AOTF. In order to accommodate for a range of velocities and different levels of brightness, the ``off'' pulse duration can be varied between 0.25 and 10 milliseconds and the ``on'' pulse duration between 15 and 200 microseconds.

One individual image frame or a sequence of up to 30 images of the flow can be captured with a CCD camera, mounted perpendicular to the light sheet and connected to a computer via a frame grabber. The camera requires that the laser produces two pulses on every second frame in order to achieve optimum image quality.

The entire system is synchronised using the composite video output signal extracted from the camera, which represents the input of the control circuit.

The schematic connections of the system components are shown in Fig. 3

Fig. 3 Schematic diagram of the PIV system interconnection

The fibre optic-lenses arrangement and the CCD camera are mounted on traverses, which can scan along the model. They allow quick positioning of the light sheet in the flow's region of interest and easy alignment of the camera.

IMAGE PROCESSING

The resolution of the captured images depends on the sensor size and the area covered. In this case, the sensor was 576x768 pixels and the flow area of approximately 1200 mm$^2$ , giving a resolution of 37.5 pixel/micron. The accuracy to which velocity data are extracted is a function of the image resolution and the pulse separation, in other words the distance travelled by a particle between the two pulses.

Two major techniques can be employed to solve PIV images. Firstly, the particle pairing method, in which individual particle images are identified, centred with sub-pixel accuracy and finally paired. Specialised image processing software, developed at University of Warwick, is employed to extract the particle positions and calculate the instantaneous velocity vectors\cite{eu96}. This technique applies mostly to flows with a low seeding concentration. The data structure obtained represents the instantaneous velocity of the field, sampled in sparse and randomly distributed points. In order to be displayed as a continuous map, a further interpolation step is necessary. The interpolation can be performed either on the nodes of a regular grid superimposed on the data, or linearly between the existent points, which are connected using a Delaunay triangulation.

The second technique, applicable to flows with medium and high seeding density, employs an autocorrelation performed on a limited size cell of the image, typically 32x32 or 64x64 pixels. The autocorrelation function will have a DC peak and also a first order peak which corresponds to the average displacement of all the particles present in the cell. The software used in this case is a commercial package named Insight. In this case, the data structure is regular. If the size of the cell is smaller than the flow feature size, instantaneous velocities are obtained.

RESULTS AND DISCUSSION

A sequence of up to 30 images of the flow can be captured continuously. The image presented in Fig. 4. is an illustration of the visualisation possibilities offered by the system.

Fig. 4. Flow visualisation digital image of the coolant passageway obtained by continuous illumination

The qualitative observation of the flow shows regions of laminar and turbulent flow and also highlights large flow features such as turning, recirculation and stagnation. Thus, the technique offers a rapid diagnostic of the flow and facilitates the identification of low velocity regions, in which a decrease in the cooling efficiency may occur. These regions can subsequently be removed by a better design of the duct shape.

As previously mentioned, the set-up gives the possibility of capturing PIV images which carry velocity information. An example of a PIV image is given in Fig. 5.

The image was solved using the particle pairing technique and the instantaneous velocity field is shown in Fig. 6.

Fig. 5 PIV digital image of the coolant passageway obtained by double pulsed illumination

Fig. 6 Instantaneous velocity map of the flow in the coolant passageway obtained from the PIV image

At the present stage, the solution has an ambiguity of 180$^ in the flow direction. However, the ``live'' flow visualisation eliminates this ambiguity. In future, this ambiguity will be removed with the use of a double frame, cross-correlation camera, which can capture each of the pulses onto two separate, consecutive frames.

The image has been analysed using the two processing methods described previously, giving both spatially averaged and instantaneous velocity vectors. The magnitude of the velocity vectors extracted from the images varies between 0.2 and 1 m/sec. The position of the particles can be estimated to a precision of 0.2 pixels.

Some small areas of the flow field were obscured by reflected stray light from the model and could not be solved. In other areas, the lack of double exposures caused the failure of the algorithms to compute velocity vectors, which suggested a three dimensional movement or high turbulence.

CONCLUSIONS

The paper described the experimental implementation of a flow visualisation and velocity measurement system, applicable to a whole range of car engine components.

The great advantage offered by the system as a whole, and especially by the manufacturing of transparent models, is the ability to carry out optical analysis prior to the construction of a real prototype. The information obtained from the flow analysis could be used interactively for improving the design of the system.

Another advantage of this system is the high accuracy (2-5\%) of the velocity data. This makes it suitable for an accurate comparison with CFD calculations. Moreover, considering the speed at which images can be captured and analysed (of the order of seconds), relative to the time required for a simulation of a flow region with complicated boundaries (of the order of hours), this experimental method has the potential to become a routine practice in design testing.

REFERENCES

1. "Lecture notes in flow measurement techniques",von Karman Institute for Fluid Dynamics, Rhode Saint Genese, Belgium, 1991-1995
2. R. J. Adrian, "Particle-imaging techniques for experimental fluid mechanics", Annual Review of Fluid Mechanics, Vol.23, pp.261-304, 1991
3. J. Kompenhans, M. Raffel, A. Vogt and M.Fischer, "Aerodynamic investigations in low and high speed wind tunnels by means of particle image velocimetry", Proceedings of the 15th ICIASF, Vol.46, 1993
4. C. Towers, P. J. Bryanston-Cross and T. R. Judge, "The application of PIV to large scale transonic wind tunnels", Laser and Optics Technology, vol.23, pp.289-296, 1991
5. P. J. Bryanston-Cross and A. H. Epstein, "The application of sub-micron particle visualisation for PIV (Particle Image Velocimetry) at transonic and supersonic speeds", Progress in Aerospace Science, vol.27, pp.237-265, 1990
6. P. J. Bryanston-Cross, D. Towers, C. Towers and T. R. Judge, "The Application of PIV in a short duration transonic annular turbine cascade", Journal of Turbomachinery, vol.114, pp.504-510, 1992
7. I. Grant and G. H. Smith, "Modern developments in Particle Image Velocimetry", Optics and Lasers in Engineering, vol.9, pp.245-264, 1988
8. G. C. Calvert, "Flow visualisation", Rapid News, Journal of Rapid Prototyping and Tooling Consortium, University of Warwick, 1994
9. G. C. Calvert, C. M. E. Driver and J. A. McDonald, "Rapid experimental analysis of flow", Time Compression Technologies Conference, Gaydon, UK, 1996
10. J. Redfern, "Measurement of thermal effect",British Society of Strain Measurement Annual Conference - Automated Strain Measurement and Analysis, University of Sheffield, UK, 1995
11. D. D. Udrea and P. J. Bryanston-Cross and M. Funes-Gallanzi and W. K. Lee, "High accuracy processing algorithms for particle centre estimation in low seeding density PIV", Optics and Laser technology, vol.28, pp.389-396, 1996

[ Home ] [ Up ] [ Current Research ] [ Prior Research ] [ OEL Members ] [ Courses ] [ Feedback ]

This page © 2001 Optical Engineering Laboratory, University of Warwick.
Questions about the OEL and related courses should be directed to Professor Peter Bryanston-Cross
Any questions or comments about this web site should be directed to the OEL Webmaster
Page last modified: November 26, 2002