Application of Laser Doppler Anemometry and Infrared Thermograph Methods for Measurements of Fluid Flow in a Small Transonic Channel
This Chapter presents a comparison of laser Doppler anemometer and Pitot-static sensing techniques for the measurement of the mean velocity and the turbulence properties in flows in a small transonic channel. The work deals with turbine blade cooling. The measurements have been done in a turbulent boundary layer which develops in jets/cross flow interaction. Comparison of precision measurement results between the two methods were made by conducting a series of tests and experiments, where the influence of the seeding over the measurements was shown. The mean and standard deviation estimated from the data are subject to both systematic and statistical uncertainty, and an analysis of these uncertainties is given. Another method of optical processing was investigated and used for stationary measurements of the temperature and the heat flux densities. This method consists of an application of an infrared thermograph technique. Conclusions are presented by the analysis of comparison results of wall temperature measurements with those taken from located thermocouples sensing.
Turbine airfoil metal temperatures in many current high temperature gas turbines are maintained at acceptable levels by means of film cooling. In order to predict the interaction behaviour between the coolant and the mainstream, and the exterior temperature and convection heat flux distribution on cooled turbine blade, a series of measurements was conducted in the closed rectangular cross section of a transonic wind tunnel. A detailed investigation based on a number of tests have been made on two-dimensional film cooling and have been concerned with the determination of velocity, temperature and pressure fields, and the determination of wall temperature. The results of this investigation are presented in Dizene et al. (1) and (2).
The present study is concerned with the measurement techniques that would lead to predictions of both the interaction phenomena in the flow field and the heat-transfer coeffcients in the neighbourhood and the downstream of a single row of injection holes with air as the film coolant and with a mainstream of air. Although a number of different geometries for the injection holes are possible, a system is chosen that approximates one used in many applications. This is a row of circular holes inclined at 45 degrees to the surface and spaced apart, centre to centre, by three diameters. Figure 2.1, which shows the complete test apparatus, gives an indication of this geometry in part of the figure. In order to study film cooling problem from a single slot, film cooling from a single row of holes, or from other system of injection, it requires a source of high temperature for practical exposure times. To simplify the test set, and to save more energy, the thermal problem was inverted. Hence, the problem remains like a cross flow study developed on a heated flat plate through which warm air was injected in the ambient mainstream. In both problems, real film cooling or the inverted thermal problem, the plate is taken at intermediary temperatures and the study is not modified as the Richardson number, which is defined as the gravitational forces to inertia forces ratio, remains negligible.
The need of several measurements to take them around and behind holes, velocity and turbulence intensity were measured by a one-component laser anemometer system. The flow field temperature was measured using a thermocouple probe, and the wall temperature distribution was measured using an infrared technique, developed by Dorignac (3).
The experiments were conducted in the University of Poitiers Aerodynamic study laboratory. All the measurements were made in rectangular cross section transonic wind tunnel. The dimensions of the tunnel test section are 40 mm high, 80 mm wide, and 600 mm long. The side walls were set to diverge slightly to maintain a uniform free stream speed of the air in the tunnel. The mainstream velocity and the boundary layer profiles were determined using a total and static pressure probe, again with static wall taps. With a normal free stream speed of 235 m/s, the boundary layer displacement thickness at the upstream edge of the injection holes was about 15 mm. The Reynolds number based on this velocity and diameter D (5 mm) was 0.78 × 105. Air was injected through five tubes spaced three diameters apart across the span. The tubes are long enough to assure fully developed turbulent flow at the exit in the absence of a mainstream flow. The flow inside the wind tunnel is developed by a supersonic ejector placed downstream of the sonic throat, as shown in Fig. 2.1. The overall flow was determined by measuring the pressure drop across an orifice plate. The injection abscise (X = 0) is located at 380 mm from the tunnel test section inlet. The dimensions of the measurement surface are X = 20 D long and Z = −1.5 D wide. The thermal test conditions are inverted by using a wall temperature of Tw = 313 K which is lower than the jets temperature Tgj = 333 K but above than mainstream temperature Tge = 286 K. Hence, a temperature difference between main and coolant flows of approximately 45 °C was obtained by heating the coolant approximately 40 °C above room temperature. The test plate was 600 mm long and was formed in the lower wall of the test section. The lower surface wall is maintained at a constant temperature by using a heated water circulation system. Seven thermocouples are embedded in the test plate at Y = 0.8 mm, and these measure and control the wall temperature and are located along the axe of symmetry (Z = 0). In these test conditions and as previously discussed, we will observe a heat flux from the wall to the flow. Then, in order to obtain an optimal cooling effectiveness, the heat flux will be negative and therefore the wall will be heated by jets.
The flow field characteristics and wall surface flow studies were realized by taking some velocity and temperature measurements. Velocity and turbulence were measured using a one-component laser anemometer system. The magnitude of the velocity vector was then estimated and the flow angle determined. The temperature above the plate was measured using chromel alumel thermocouples having a recovery factor equal to 0.88. Detailed measurements were performed using cross flow momentum flux ranging from about 3–5 times the jets momentum fluxes. For data processing, the output voltages of the sensors were stored numerically on a computer. An amplifier multiplexed intelligent card for data acquisition collects the output voltages.
In numerous flow investigations, the variables required are the components of the mean velocity vector and the turbulence intensities. In our configuration, further information on the flow which is included in the Reynolds shear stresses and distribution excesses, is often of little interest in these investigations. The longitudinal and vertical mean velocity distributions were measured using a one-component LA system with its optical arrangement shown in Fig. 2.2. The beam splitter device is of an auto corrector model. The following numbers indicate : 1 the beam expander, 2 the dicrotic beam splitter, 3 deviation prism, 4 the Brag cell, 5 the collimator lens. The elements 2, 3, and 4 are interdependent and mounted on a pivoting arbour. The collimator lens assure, the beam convergence. Because of the relatively high compressible flow velocity, all measurements were made by a forward scattering set-up. The transmitter–receiver optical displacement is realized by an electrical motor moving step by step with a resolution of 0.1 mm. Fringe defilement is realized by Brag cell at a frequency of 40 MHz.
The seeding method consists of an injection into the mainstream of pulverized vegetable oil particles 1 μm in diameter size. The use of the seed injection into the flow field in order to enable LA measurements can result in the accumulation of seed particles on the windows inner surfaces. This deposition is due to large particles which do not follow the flow and are therefore scotched on the surface of the window. The photo multiplier can be saturated because of reflected light entering the collection optics. Reflections from window surfaces can be reduced by using anti-reflection coatings on the window. The mean velocity components u and v are respectively measured. The measurement direction is changed when the fringes are oriented to measure the other component. Data processing is performed using a TSI processor 1990 model. The output signal is received on an intelligent CTM 05 interface card, manufactured by Metrabyte located in a computer. The uncertainties in the measured velocity were approximately ±4 m/s.
A second measurement technique was developed to obtain detailed film effectiveness data using a scanning camera. Data for this investigation were obtained in the same open circuit transonic wind tunnel. A view of the tunnel installation in the region of the test model is presented in Fig. 2.3. The surface temperature and heat fluxes on the flat plate model were determined using an AGEMA infrared scanning camera SWB 880. This camera had a depth of field of approximately 5 cm when focused on the test section and 35 cm from the camera. The infrared viewed the test surface through a scuttle of vinyl chloride which is used as an infrared window in the test section up wall. The camera was calibrated and the emissivity coefficient obtained was 0.87. The test surface of the model was coated with high emissivity flat-black paint. The thermocouple measurements provided accurate temperature measurements at seven locations and the infrared scanning camera provided isotherm contours at selected temperature levels which would have been possible only with a multitude of thermocouples.
For some coolant flow rates, four separate isothermal levels were mapped along the X direction with a 250 × 250 pixel resolution. In the span wise direction which the length covered is about X = 40 D. The flux density is deduced from an application of a numerical method developed by Dorignac (3), based on a stationary conduction model. The validation method is obtained by comparing the isotherm contour value with the value resulting from thermocouple measurement at the same position. The measured temperature is influenced by numerous parameters like emissivity coefficient and radiation issued from the surroundings. The data are presented in a plane view of the flat plate with the coolant holes. The hole position on the isotherm display is difficult. This position was realized using high temperature variations. Then the isotherm contours were positioned with an uncertainty of ±0.26 mm. The uncertainties in measured temperatures taken by the scaning camera were approximately 10 per cent (3).
The trade-offs between seeding the entire flow field, using a point source of seed, or not seeding are briefly discussed below. When designing a laser anemometer experiment, we must decide whether to seed the entire flow field and jet flow (full coverage) or not to use seed in the jet flow (without injection). Both methods are used in our experiments and compared to see by which method the signal strength can be improved relative to the background noise. The method chosen is dependent on the flow field characteristics. The seeder used has been designed in the aerodynamic studies laboratory (LEA in FRANCE) which uses natural vegetable oil. This seed material is formed by pulverizing a particle vegetable oil which is injected into the flow field. The seeder is placed far enough from the upstream to enable the decay of the wake shed before the measurement point is reached.
The results of these two methods are shown in Fig. 2.4. When not using seed, the light reflected from solid surfaces is orders of magnitude higher in intensity than the light scattered by the particles which are naturally present in atmospheric air, so this dominates the PMT signal, i.e. the signal to noise ratio of the PMT signal remains minimum or may drop to zero. Seed particles are injected locally into the upstream flow, thus only a comparatively few of the particles may penetrate the wake flow where measurements are being taken So, this is not sufficient to yield adequate data rates and we observe a low turbulence intensity in all measurement planes. When seed is used (full coverage), the signal to noise ratio remains constant, so we observe a high turbulence intensity in all planes than without seeding and validate reference ratio remains 100 per cent.
Our optical configuration has been constructed for measuring one velocity component. The simplest method of measuring velocity magnitude and flow angle is to acquire measurements at three different fringe orientations. The statistical error, measured with a single component system is not greater than that resulting from a two-component system as was described in (1). The magnitude of the turbulence components u′ and v′ are also determined with measurements taken in the X and Y co-ordinate directions, and compared in measurement planes located without distortions, with that resulting from a two-component system (4) in Fig. 2.5. We observe satisfactory agreement between both measurement results.
In addition, the magnitude of velocity vector V and flow angle α measured with a one-component laser anemometer system are presented and discussed in Fig. 2.5. The measurement velocity components are made along the symmetrical axe (Z/D = 0) and in lateral positions located at Z/D = 6 and 20. Results are presented respectively in Figs 2.6(a) and 2.6(b) and compared with that resulting from a single Pitot probe tube. Great differences in magnitude are observed in the symmetry plane until X/D = 2, i.e. where the flow angle remains large (about α = 15 degrees).
We observe here that the velocity vectors resulting from a laser anemometer system measurement are usually much larger in magnitude than the velocity vector resulting from Pitot tube measurements. Use of Pitot tube probe results in two greater limitations. The first one is its sensitivity to accurate estimations of mean velocity vector when the flow angle is greater than 15 degrees. The measurement error is estimated in this way at about 20 per cent. Figure 2.6(b) illustrates great differences particularly in symmetry plane (Z/D = 0). For the downstream flow field (Z/D = 20) when the flow angle is small, the values resulting from the two measurement techniques are in good agreement. The second one is obviously due to acoustic effects which have been present outside and more inside the turbulent boundary layer. From the above results and taking into account the order of magnitude of the instantaneous flow angle, it appears clearly that for the flow configuration of interest here, the scale pressure mean can not be measured with an acceptable accuracy by using a classical Pitot static and Pitot total tube.
The brievely flow description obtained with a one and two-component velocity measurements and presented here show that the statistical error in velocity measured with a single component system will, in most applications, be to date. Single channel laser anemometer systems have both aerodynamic and turbomachinary applications. The advantages of a single channel system are its simplicity and the fact that the available laser power is concentrated into a single fringe system.
In studying the two-dimensional film cooling from a single row of holes, a convenient means of analysing the problem has been to consider the wall temperature and the heat transfer coefficient as separate quantities to be determined. The wall temperature can be put in a convenient dimensionless form θw as defined previously. As for aerodynamic fields, the thermal quantities have been determined in regions around holes, behind the holes, and downstream the holes. The measured wall temperature evolution is shown in Fig. 2.7. The figures in the span wise planes (X, Z) are given in order to appreciate the phenomena in the Y direction.
The plane located at position X = 0 and at very close station measurements, the thermal effect, induced by the warm jet exit, may be seen. Downstream, points out the mixing and the thermal diffusion of the jets. Wall temperatures were measured in planes with the thermocouple probe located at positions X/D = 1; 2 and 10 along the injection tubes and in the mainstream. The profiles close the wall (Y/D < 1) are only presented in Fig. 2.8. These results are compared to heat gradients from the surface wall to the turbulent boundary layer that are indicated by the infrared camera. The temperature values indicated by the camera are below the point value which is close to the wall and the value which is measured by using a thermocouple probe. Hence, the slope is considered positive and we can think there are some points located near the wall, which are not measured by the thermocouple probe position, where the temperature profiles presents a minimum value. This phenomena is an unfavourable effect on the heat transfer between the jets and the wall.
A series of tests and experiments in a small laboratory transonic wind tunnel has been conducted using a laser Doppler anemometer system and an infrared thermograph measurement technique. The method of seeding applied to the LA measurement system is presented, along with the results from classical, static, and total pressure tube measurements to estimate the velocity in a turbulent boundary layer and both methods are compared. The results show that for the flow configuration considered here, the classical method does not allow measurement of the in-stream mean pressure Pitot tube with an acceptable accuracy. Some results issued from the method of a one-component LA system at two different fringe orientations are compared with that from a two-component LA system for mean and fluctuation velocity measurements. Statistical error in the velocity measured with the one-component LA system is acceptable.
(1) Dizene, R., Charbonnier, J. M., Dorignac, E., and Leblanc, R. Etude expérimentale d'une interaction de jets obliques avec un écoulement transversal compressible. I. Effets de la compressibilité en régime subsonique sur les champs aérothermiques. International Journal of Thermal Science, March 2000, vol. 39 N°3, pp 390–403.
(2) Dizene, R., Dorignac, E., Charbonnier, J. M., and Leblanc, R. Etude expérimentale d'une interaction de jets obliques avec un écoulement transversal compressible. II. Effets du taux d'injection sur les transferts thermiques. International Journal of Thermal Science, May 2000, vol. 39 N°5, pp 571–581.
(3) Dorignac, E. Contribution à l'étude de la convection forcée sur une plaque en présence de jets pariétaux dans un écoulement subsonique. Thèse de Doctorat. Université de Poitiers, 1990.
(4) Bousgarbies, J. L., Foucault, E., Vuillerme, J. J., and Dirignac, E. Etude de l'interaction jets/écoulement en paroi plane. Refroidissement des aubes de turbines par jets – rapport final, contrat DRET. Décembre 1991.
Blair, M. F. and Lander, R. D. New techniques for measuring film cooling effectiveness. Journal of Heat Transfer, November 1975, pp 539–543.
Goldstein, R. J. and Taylor, J. R. Mass transfer in the neighborhood of jets entering a crossflow. Journal of Heat Transfer, November 1982, vol. 104. pp 715–721.
The authors wish to express their thanks to DRET (DGA) institution for their financial help and Mr Henry Garem for his contribution with the conduction of all experiments. The authors wish to express their thanks to Professor David Zeitoun from IUSTI (Marseille, France) for his interest in this work.
Department of Mechanical Engineering, USTHB University, Algiers, Algeria
E Dorignac and R Leblanc
Laboratoire D'etudes Thermiques, ENSMA, Futuroscope, France
J M Charbonnier
CNES, Toulouse, France