Experimental study of mid-ir laser beam wander close to a jet engine exhaust

Transcription

1 Experimental study of mid-ir laser beam wander close to a jet engine exhaust Markus Henriksson *a,b, Lars Sjöqvist a and Ove Gustafsson a a Dept. of Laser Systems, Swedish Defence Research Agency, FOI, P.O. Box 1165, SE , Linköping, Sweden. b Department of Physics, Royal Institute of Technology, Stockholm, Sweden ABSTRACT An increasing interest in lasers placed on aircrafts for active countermeasures and active imaging is observed. There remain unsolved issues regarding the propagation effects close to the jet engine exhaust and the possibilities of compensating them with adaptive optics. Laser beam propagation experiments parallel to the exhaust of a downscaled jet engine test rig have been performed. The experiments were carried out with nanosecond laser pulses at 1.6 and 3.5 µm wavelength. The laser spots were projected on a screen and the centroid motion were imaged by cameras. Root mean square magnitudes of the beam wander between 5 and 15 µrad were observed for different engine conditions and geometries. The 3.5 µm system had a frame rate of 67 Hz and could partly resolve the time variation of the beam wander. A correlation time (5 %) of 3.5 ms was observed for the beam wander. Deflections of several hundred µrad due to the average gradients in temperature and pressure were also found when the engine was turned on. In addition to beam wander intensity scintillations has been studied. Keywords: Laser, beam wander, countermeasure, jet engine. 1. INTRODUCTION Refractive index variations in air, commonly discussed as turbulence, due to variations in temperature, pressure and gas composition will cause refraction of a laser beam. These time-varying disturbances result in effects such as beam wander, beam broadening and scintillations. For atmospheric propagation there is a developed theory for predicting the order of magnitude and temporal statistics of the disturbances. In recent years there has been an increasing interest in mounting lasers on aircrafts for applications such as laser based countermeasure systems (DIRCM), active imaging and optical communication. Additional disturbances of the air originating from the aero-optical effects close to the fuselage of the aircraft and by the engine efflux then need to be considered. The turbulence caused by the hot exhaust gases from a jet engine will be much stronger than normal atmospheric turbulence and may show quite different spatial and temporal statistics. To be able to predict the performance of aircraft based laser systems at different pointing directions and the usefulness of adaptive optics correction techniques it is important to characterize the turbulence caused by the jet engine. Experimental studies on laser beam propagation through jet engine plumes have been reported by several groups. In the early study by Hogge and Visinski laser beam propagation parallel to an engine plume was used to estimate the averaged structure parameter describing the turbulence strength 1. More recently laser beam propagation effects including beam broadening and wandering in strong turbulence with very high value of the structure parameter originating from the exhaust of a jet plume from a DeHavilland Vampire aircraft were studied 2. The beam spreading and centroid motion of visible and mid-ir laser beams propagating through a jet engine plume have been studied by using cameras to view and register perturbation effects 3. Laser beam propagation through jet engine plumes have been studied comprehensively by Sirazetdinov and coworkes 5,6,7,8. They studied the influence from the plume on parameters such as beam wander and beam broadening for different laser wavelengths and propagation geometries. A dedicated experimental setup was described for studying both near- and far-field beam perturbation effects. Multi wavelength * mahe@foi.se, Phone: , Fax: Technologies for Optical Countermeasures III, edited by David H. Titterton, Proc. of SPIE Vol. 6397, 63979, (26) X/6/$15 doi: / Proc. of SPIE Vol

2 experiments using mid-ir lasers and fast imaging cameras were reported recently 9. Our group has reported measurements at 1.6 µm in a previous study 1. In this work laser beam propagation experiments close to a downscaled jet engine plume were performed. The propagation paths were chosen to be parallel to the flow axis of the engine. The experiments involved the use of two wavelengths, registrations of beam perturbations such as beam wander, scintillations and analyzing temporal variations. Different engine conditions with similar drag were used. Finally a summary of the experimental trends we found is presented and a short discussion of the impact on laser systems in applications such as DIRCM and active imaging is given. 2. EXPERIMENTAL SETUP The experiments were performed using a downscaled model jet engine located at Volvo Aero in Trollhättan, Sweden. The engine is designed to provide a characteristic exhaust in comparison to full scale engines. Pressurized air is provided to the burn chamber and as a free streaming shear layer to simulate flight conditions. Air and fuel flows as well as the engine geometry can be varied to investigate different phenomena. The setup also includes sensors to monitor engine performance during the run. The engine flow axis was situated approximately 1.5 m above the ground. During the experiments two different engine nozzles, one circular and one elliptical, were used. The elliptical nozzle was also rotated to investigate the performance along the different axis. Each engine run lasted typically 2 to 3 minutes during which only the amount of free-streaming air varied. In total seven runs with different engine parameters were used in the analysis. Two different laser and camera setups were used to study the laser beam propagation along the jet engine exhaust at 1.57 and 3.5 µm. The 1.57 µm setup used a flashlamp pumped Nd:YAG laser with an optical parametrical oscillator (OPO) for wavelength conversion (Big Sky Lasers/Quantel Brilliant Ultra OPO). The pulse repetition frequency was 2 Hz, maximum pulse energy 8 mj and the pulse duration approximately 8 ns. The laser was equipped with an x1 telescope (CVI HEBX-4.-1X) to enlarge the beam diameter. The resulting divergence was approximately 1.9 mrad (full angle). The laser was placed behind the engine and the beam was projected on a sandblasted aluminum screen to ensure diffuse reflection. The screen was placed on a fence and stabilized to minimize movement. The laser spot on the screen was imaged using an InGaAs camera (Sensors Unlimited Inc. SU321.7RT). A narrowband interference filter was used to limit the amount of background radiation collected. The camera was synchronized to the laser to ensure exposure of each frame with one laser pulse. The 3.5 µm setup used a diode pumped Nd:YVO4 laser (Azura Mesa AC IR) running at 1 khz repetition rate and an OPO based on Periodically Poled Lithium Niobate (PPLN) to convert the wavelength. The pulse energy was around 2 µj and the pulse length 3 ns. The laser divergence was approximately 1.5 mrad (full angle). The laser was projected on the same screen as the 1.57 µm system. The imaging camera was a cooled cadmium mercury telluride array (AEG Infrarot-Module GMBH, Germany). To reduce the thermal background a filter transparent between 3.45 and 4.15 µm was used. The camera frame rate was approximately 67 Hz. The integration time of the sensor was 1 µs to sample one pulse with each frame. When checking the amplitude in the images no evidence of double exposure has been observed. The overall setup can be found in Figure 1. Proc. of SPIE Vol

3 MCT camera InGaAs camera 3.5 µm laser Engine 1.57 µm laser Figure 1. The experimental setup. Framework structure Screen The lasers were placed nine meters behind the engine nozzle and the distance from the nozzle to the screen was 33 m. The laser propagation paths were located above the engine plume to minimize ground effects. Due to mechanical constraints they were slightly diagonal to the engine flow axis. Two different positions relative to the engine axis were investigated for each wavelength. The distances from engine axis to laser path varied from.4 to about 1 m. Eight meters in front of the nozzle a framework structure to dampen the engine noise was located. The major contributions from large scale turbulence cells on the laser propagation are thus believed to originate in this 8 meter distance. The laser beams were directed through the holes in the framework and care was taken to avoid clipping. The cameras were placed in an off-axis position relative to the engine axis and imaged the screen that was placed with the normal at a small angle to the laser path. For the mid-ir camera each pixel corresponded to 2.67 mm or 81 µrad deviation in the vertical direction and 3.47 mm or 15 µrad deviation in the horizontal direction, with the angles calculated from the engine nozzle. For the InGaAs camera each pixel corresponded to 4.88 mm or 148 µrad vertically and 6.47 mm or 196 µrad horizontally during the first day of measurements. The second day a different camera position closer to the screen was used where each pixel corresponded to 2.28 mm or 69 µrad horizontally and 2.41 mm or 73 µrad vertically, respectively. 3. EXPERIMENTAL RESULTS To maximize the signal to noise ratio background images were registered and subtracted before the analysis. The images were also cropped to reduce the influence of noise. During the first day of measurements the 1.57 µm laser beam was situated relatively far away from the engine flow axis and as depicted in Figure 2, left side, the beam profile looks almost unaffected. Before the second day of measurements the laser was moved closer to the engine flow axis and the turbulence effects on the laser beam profile were larger as depicted in Figure 2, right side. The 3.5 µm setup showed similar high distortions during both days even though the laser position was adjusted slightly between the different measurements occasions. Examples of images of the beam with the engine off (left) and on (right) are shown in Figure 3. Recordings were made as image sequences, normally containing 5 images for the 1.57 µm system and 9 images for the 3.5 µm system. Since the sequences are long on the time scale of the turbulence all conditions yield smooth intensity distributions when averaging in time. Due to the fact that the beams are relatively wide and the distance from the engine to the screen is short no significant broadening of the beam is observed. Increases in beam divergence of up to 1 % were calculated but since the beam width is very difficult to define in noisy images the results are not accurate enough to provide definite conclusions. Proc. of SPIE Vol

4 1 1 Vertical position [mm] 5 5 Vertical position [mm] Horizontal position [mm] Horizontal position [mm] Figure 2. Beam images from the 1.57 µm system with the engine on, on the left from the first day and on the right from the second day Vertical position [mm] Vertical position [mm] Horizontal position [mm] Horizontal position [mm] Figure 3. Beam images from the 3.5 µm system, on the left side with the engine off and on the right side with the engine on. Calculations of instantaneous beam position were performed with a Fourier transform based centroid finding algorithm since this method is believed to be less sensitive to noise and dead pixels than the standard first moment centroid calculation method 11. The maximum deflections of the beam from the average position, when assuming that all refractive power is concentrated at the nozzle, is around 4 µrad. In Figure 4 the pattern traced by the 3.5 µm beam centroid is shown and compared to the reference case without the engine running. A large increase in the centroid motion is observed when the engine is turned on. The beam wander with the engine turned off is believed to mainly be a result of laser pointing jitter. On the left side of Figure 5 a time series of the horizontal deflection of the 3.5 µm beam is shown. The beam position follows a Gaussian distribution as exemplified on the right side of Figure 5 for one of the experimental data sets recorded with the 1.57 µm setup. Proc. of SPIE Vol

5 4 4 Vertical deviation from mean [µrad] Vertical deviation from mean [µrad] Horizontal deviation from mean [µrad] Figure 4. The pattern traced by the 3.5 µm beam centroid with the engine off (left) and on (right) Horizontal deviation form mean [µrad] 3 3 Horizontal deflection [µrad] Frequency Time [s] Horizontal beam deviation [µrad] Figure 5. Time series of the horizontal deviation of the 3.5 µm beam (left) and histogram of the horizontal deviation for the same measurement (right). The dotted line in the histogram is a fit to a Gaussian distribution. As turbulence is a random process the exact beam wander cannot be predicted, but many conclusions can be drawn from the statistics of the beam motion. In Figure 6 the root mean square (rms) beam wanders in horizontal and vertical directions for the 1.6 (left) and 3.5 µm (right) beams, respectively, are shown. In all cases an increase of between 3 and 1 times is observed with the engine on compared to the reference measurements with the engine off. The factor two differences in magnitude between individual measurements can partly be explained by the fact that different engine configurations were used, but even within the sets that are shown with the same marker some variation can be observed. In the 1.6 µm data an asymmetry where the measurements from day 2 (diamonds, dots, upwards triangles and downwards triangles) show more movement in the horizontal direction than in the vertical direction. During day 1, on the other hand, (stars, circles and squares) the movement was more symmetrical. There is, however, no change in the magnitude, even though the beam suffered more from small scale turbulence during the second day as seen from the beam images (Figure 2). The asymmetry is attributed to the fact that the beam passed almost straight on top of the engine axis during the second day, whereas it passed with almost equal horizontal and vertical distances to the engine plume on the first day. For the 3.5 µm setup the laser beam propagated with a larger horizontal than vertical distance to the flow axis during the second day and the vertical beam wander was then slightly larger than in the horizontal direction (diamonds and dots). This could be interpreted as an indication that the magnitude of the beam wander is Proc. of SPIE Vol

6 larger in the tangential direction than in the radial direction relative to the engine axis, but additional measurements have to be performed before any certain conclusions may be drawn Vertical rms beam wander [µrad] Vertical rms beam wander [µrad] Horizontal rms beam wander [µrad] Horizontal rms beam wander [µrad] Figure 6. Rms beam deviation for the 1.6 µm (left) and 3.5 µm (right) systems. The points in the lower left corners are reference measurements while the different symbols show different engine conditions and propagation geometries. The dotted lines are symmetry lines. With the 67 Hz sampling rate used in the 3.5 µm setup it seem like the variation in time is at least partly resolved when studying Figure 5 (left graph). Thus it is interesting to study the time dynamics of the beam centroid motion. When plotting the power spectral density on a logarithmic scale a high frequency roll-off of f horizontally and f vertically is found, see Figure 7, left side. The curves in the figure are averages of all relevant measurements. For atmospheric turbulence an f -8/3 dependence would have been expected 12. The result of calculating the autocorrelation of the beam wander and averaging over all measurements is shown on the right side in Figure 7. The measurements with the engine on show a 5 % correlation time of 3.5 ms, which is slightly more than twice the time between two consecutive samples. The reference measurements show no correlation, confirming that the measured correlation is caused by the engine exhaust. A similar calculation for the 1.6 µm system as expected show no correlation since the sample rate is too low. Power spectral density [A.U] Vertical Horizontal Correlation, normalized Horizontal, motor on Vertical, motor on Horizontal, motor off Vertical, motor off Frequency [Hz] Time [ms] Figure 7. Left: Power spectral density of the beam wander for the 3.5 µm system in horizontal and vertical direction. Right: Autocorrelation of the deviation from the mean position for the 3.5 µm system. Proc. of SPIE Vol

7 In addition to the random beam wander due to turbulence cells a constant shift of the beam position was observed. This is very evident on the left side in Figure 8 where data registered with the 1.6 µm setup during engine ignition is shown. The movement of the beam is caused by a gradient in the index of refraction between the hot engine plume and the surrounding cold air. The refraction will thus always be towards the engine axis. Vertical deflection [µrad] Vertical deflection [µrad] Time [s] Horizontal deflection [µrad] Figure 8. Left: Time series of the vertical deflection of the 1.6 µm beam during start of the jet engine. The shift due to the gradient between the plume and the free air is clearly seen as well as the increase in random movement. Right: Mean deflections of the 3.5 µm beam compared to reference positions. Each point is the average of the 9 frames in one measurement. The different markers show different engine conditions and propagation geometries. The variation within each set is partly attributed to different amounts of free streaming air. Comparisons of the beam positions with the engine on and off show large differences (both between and within experimental data sets). The deflections of the 3.5 µm beam are shown on the right in Figure 8 with each kind of markers corresponding to one engine run. Variation between the sets can be explained by the different nozzles and other changes to the engine and by the two different laser positions. Within the sets some grouping of the data points related to the amount of free streaming air can be observed, but there are still significant variations that need to be explained Intensity [A.U.] 1 Intensity [A.U.] Time [s] Time [s] Figure 9. Time series of the power in an aperture with the engine off (left) and on (right). The example is from the second day of 1.57 µm measurements. A traditional way of characterizing turbulence is to measure the variation of the power in an aperture. By introducing a software aperture in the images this can be simulated in the experimental data. Since the aperture was fixed with the center at the average beam position a combination of scintillation and beam wander effects was measured. Examples of Proc. of SPIE Vol

8 the variation over time are shown in Figure 9 for the 1.57 µm system. The figure compares the variation with the engine on to the reference measurement with the engine off. In all cases except the first day of 1.57 µm measurements there is a significant difference between the on and off results. The fact that the scintillations caused by the engine are small during the first day is no surprise since the beam profile is very smooth (Figure 5). No correlation over time was found for the 3.5 µm system and the conclusion is that the correlation time attributed to the scintillations is below about 1 ms. The statistical measure of the scintillations is the scintillation index defined as 2 2 I I 2 σ I =, (1) 2 I where I is the instantaneous intensity in the aperture. If the aperture is larger than the smallest structures in the scintillation pattern the variations will be smoothed (aperture averaging). The scintillation index as a function of aperture radius for the 3.5 µm system is shown to the left in Figure 1. Each curve in this graph corresponds to one separate measurement. The variations in the scintillation index for small apertures between different measurements during similar condition are less than a factor 3. When the aperture is larger than the turbulence cells or than the beam a rapid decrease of the scintillation is expected. The theoretical value for the large aperture roll-off in atmospheric turbulence is -7/3. Values calculated from our measurements are in the range of -1 to -2.7 with a median of -1.7 for the 3.5 µm data. For the second day of 1.6 µm measurements the values range from -1.1 to -2.7 with a median of The large spread is probably caused by effects of the beam being of the same size as the aperture. Since the theoretical calculations assume a Kolmogorov type spectrum for the index of refraction variation differences between our measured values and the -7/3 exponent were to be expected. Scintillation index Aperture radius [mm] Vertical position [mm] Horizontal position [mm] Figure 1. Left: Scintillation index as a function of aperture radius for all measurements with the 3.5 µm system. The aperture averaging gives some information on the size of the turbulence cells. Right: The scintillation index of each pixel for one measurement with the 3.5 µm system. The values in the middle of the beam are quite close to the values for a centered small aperture. A small increase is seen at the edges of the beam where the influence of the beam wander is larger. The pixels containing only camera noise have much lower variance By calculating the difference in scintillation index for a fixed aperture and an aperture that moves with the beam centroid the contribution from the beam wander may be estimated. With the 1.57 µm system during day 2 the contribution from the beam wander is 2 % for small apertures and reaches 5 % as the aperture gets to the size of the laser beam. These low numbers are of course a result of the large divergence of the laser beam. In a system with lower divergence the beam wander may be of the same order as or larger than the divergence and consequently the contribution from the beam wander will increase. Another way of illustrating the turbulence in the image is to calculate the average scintillation in each pixel. An example of the distribution of the scintillation index for the second day of 3.5 µm measurements is given in Figure 1. Comparing this image to the intensity distribution in Figure 3 it can be seen that the normalized variance is largest on Proc. of SPIE Vol

9 the edges of the beam where the gradient of the intensity is steeper. This can be explained by the beam wander component being more significant here. Ideally the average scintillation index over the beam should correspond to the scintillation with the aperture average effect over the area of one pixel. The averaged scintillation over a small aperture in the centre of the beam is very close to the scintillation measured for the total power in the same aperture. 4. DISCUSSION AND CONCLUSIONS Laser beam propagation along the exhaust plume from a jet engine has been measured for two different wavelengths, 1.57 and 3.5 µm. The data set is limited and no definite conclusions on the dependence of different parameters have been made. We have observed several tendencies that are presented here but further experiments should be carried out to confirm these observations. The engine used was a downscaled model engine and for a full scale engine larger effects would be expected. The difference of having an engine situated on the ground compared to an aircraft maneuvering in the air should also be kept in mind. The observed rms beam wander for both wavelengths are in the region of 6 to 15 µrad. The deflected angles show a Gaussian distribution and the maximum beam deflection angles are around 4 µrad. These angles are defined as if all refraction was concentrated in a plane at the engine nozzle. The real refraction will occur over a distance downstream to the engine and the angles are thus somewhat underestimated. Correlation to different engine conditions and propagation geometries was observed but no discernable trends were identified. Evidence of a slight radial asymmetry with respect to the engine axis, with larger tangential than radial beam wander, was observed. In addition to the random beam wander a constant shift of the beam position due to the gradient in the refractive index between the hot exhaust gases and the surrounding air was evident. The refraction angle was several hundred µrad and always directed towards the engine axis. The 3.5 µm system had a sampling rate of 67 Hz and thus the time variation of the beam wander could be studied. The power spectral density of the beam wander showed an exponential roll-off at high frequencies of and for the horizontal and vertical directions respectively. The 5 % correlation time of the beam centroid position was found to be 3.5 ms. Most published data on atmospheric turbulence regard the scintillation index. Due to the short propagation distance small values of the scintillation index were found. The scintillations showed a significant decrease with distance from the engine axis to the laser beam, while the beam wander was nearly constant for the same translation. The purpose of studying laser beam propagation in severe perturbed environments such as in close neighbourhood to jet engine plumes is to be able to predict performance of laser based systems located on airborne platforms. The primary application is DIRCM aiming to protect aircrafts from heat-seeking missile threats. The turbulence caused by the plume is a very complex environment for beam propagation. However, in some occasions and threat situations it might be necessary to direct the tracking system field of view and the outgoing laser beam close to air masses disturbed by the hot turbulent plume. On these occasions the performance of the system can be affected in terms of beam wander and accumulated phase distortions giving rise to intensity scintillations which affect the PIB at the target and the temporal properties of the jamming waveform. Moreover, the tracking accuracy is also affected both in the open and closed loop system configurations. The open loop system operates without optical feedback while the closed loop system, on the other hand, uses the optical signature from the target to improve tracking and optimize the jamming sequence. The possibilities to use adaptive optics to increase the amount of power on the target are greatly influenced by the temporal statistics. Our results show that tip-tilt compensation with a loop time of around 1 ms would yield a significant improvement. ACKNOWLEDGEMENTS The personnel at Volvo Aero in Trollhättan, including Anders Hellgren and coworkers, are gratefully acknowledged for the support at the engine facility. The project was supported by the Swedish armed forces. REFERENCES 1. Hogge, C. B., and W. L. Visinsky, Laser Beam Probing of Jet Exhaust Turbulence, Appl. Optics, vol. 1, pp , 1971 Proc. of SPIE Vol

10 2. Barett, J.L., and P. A. Budni, Laser beam propagation through strong turbulence, J. Appl. Phys, vol. 71, pp , Titterton, D., Measurements of the Distortion Generated in a Laser Beam s Characteristics Resulting from Passage through an Engine s Wake, RTO SET Symp., RTO MP-1, pp. 42-1, D. I. Dmitriev, Y. N. EvchankoI. V. Ivanova and V. S. Sirazetdinov, Multiframe recording of laser radiation distorted by a turbulent jet from an aircraft, J. Opt. Technol., vol. 68, pp , Evchenko, Y.N. et al., Spatial characteristic of laser radiation perturbed by a turbulent air flow, J. Opt. Technol., vol. 65, pp. 139, V.S. Sirazetdinov, D.I. Dmitriev, I.V. Ivanova and D.H. Titterton, Effect of turbo-engine jet on laser radiation. Part1. Angular spectrum of disturbed beam, Atmos. Ocean. Opt., vol. 14, pp , V.S. Sirazetdinov, D.I. Dmitriev, I.V. Ivanova and D.H. Titterton, Effect of turbo-engine jet on laser radiation. Part2. Random wandering of disturbed beam, Atmos. Ocean. Opt., vol. 14, pp , V. S. Sirazetdinov, D. I. Dmitriev, I. V. Ivanova and D. H. Titterton, Effect of turbulence intermittence on the structure of laser beams intersecting an aero-engine exhaust, Proc. SPIE, vol. 526, 1-111, W. M. Isterling, L. J. Cox, M. Dubovinski, D. H. Titterton and T. Porter, Laser Interaction with Jet Engine Induced Turbulence, 4th Australian Conf. on Laser Diagnostics in Fluid Mechanics and Combustion, Adelaide, 7-9 Dec., L. Sjöqvist, M. Henriksson and O. Gustafsson, Laser beam propagation in close vicinity to a down-scaled jet engine exhaust, Proc. SPIE, vol. 5615, pp , E. Weisshaar, G. Küveler and M. Bianda, Schnelle und genaue Methode zur Schwerpunktsfindigung in Messreihen, Photonik, vol. 4-23, pp , 23 (in german). 12. J. Dunphy and R. Kerr Turbulence effects on target illumination by laser sources: phenomenological analysis and experimental results, Appl. Opt., vol. 16, pp , Proc. of SPIE Vol