Investigation of the spray characteristics of a twin fluid atomizer using LSI and PDA

The purpose of this paper is to fully characterize an industrial twin fluid swirl atomizer through the use of laser based measurement techniques. The atomizer used in this study was manufactured by Delavan and features three air swirl inlets onto the main fluid core located in a mixing chamber within the nozzle. The two phase flow then impacts a strategically positioned impingement plate which promotes further mixing. Finally, the two phase mixture exits the nozzle via an annular passage formed using a pintle. Figure 1 illustrates the concept of this industrial atomizer.

Investigation of the S p r a y Characteristics of a T w i n Fluid A t o m i z e r Usi n g LS I a n d P D A Stephane G. Daviault and Edgar A. Matida Department of Mechanical and Aerospace Engineering Carleton University Ottawa, ON, K1B 3VS Omar B. Ramadan and Patrick M. Hughes Combustion Measurements and Kinetics Canmet Energy Ottawa, O N August 30, 2011 1 Water Introduction The purpose of this paper is to fully characterize an industrial twin fluid swirl atomizer through the use oflaserbased measurement techniques. The atomizer used in this study was manufactured by Delavan and features three airswirl inlets onto the main fluid core located in a mixing chamber within the nozzle. The two phase flowthen impacts a strategicallypositioned impingement plate which promotes further mixing. Finally, the two phase mixture exits the nozzle via an annular passage formed using a pintle. Figure 1 illustrates the concept of this industrial atomizer. Impact Plate SectionA:A Figure 1: Twin fluid swirl atomizer Optical characterization techniques can aid in the design, and predict the performance of,nozzles in in­ dustrial applications. The important characteristics required foroptimum spray performance includes the spray angle,the velocityand the droplet sizedistribu­ tion. The design of the combustion chambers, gasifiers or burners depends on the droplet trajectory, penetration rates and residence times. The symmetry of the spray is of paramount importance for manu­ facturers as an asymmetrical mass delivery results in significantly different local equivalence ratios in the burners. This paper presents the spray characteristics of a twin fluid swirl atomizer designed for use in gasifiers, burners and furnaces. This work is of great signif­ icance to nozzle manufacturers, researchers and the energy industry. With this data, itispossible to im­ prove the nozzle design, understand how the nozzle performs in a gasifier and model the interaction of the nozzle spray with the gasifier flow field. 1 Shutt-off Valve Qualitative spray characteristics such as the cone angle, spray solidity and detailed structure geometry near the nozzle exit were found using a combination of laser sheet imaging (LSI) and back-lit strobe im­ ages. The atomizer quantitative characteristics such as the velocity and size of droplets were found using a phase Doppler interferometric technique. 2 Experimental apparatus In this report, spray characteristics of the atomiza­ tion of a twin fluid water/air mixture from a swirl atomizer is performed using a combination of Laser Sheet Imaging (LSI) and Phase Doppler Anemometry (PDA). The measurements were performed in an unconfined atmospheric ambient condition. 2.1 Spray system The spray system used in this paper is illustrated in Figure 2. The swirl nozzle studied is a Delavan Figure 2: Schematic diagram ofthe atmospheric un­ A32740-7 body with tip ID 707-13. LSI was imple­ confined experimental apparatus using LSI and mented with a Ar-ion continuous wave laser which PDA had a maximum power of 2W while the PDA mea­ surements were conducted using a similar Ar-ion laser with a higher power output of 7W. the necessary information needed for the evaluation of the different flow conditions studied. 2.1.1 Imaging Equipment The camera used in this paper is a JAI- CV-M9CL Progressive scan RGB color camera composed of 3 CCD sensors (3 x 1/3) and a sensitivity of 1034 (h)x 779(v) pixels. The integration time range of the JAI camera is 33 ms to 0.02 ms. This camera is used to capture the spray angle and solidity as well as the spray structure near the nozzle exit using a magnifi­ cation lens. To ensure accurate visual representation of the spray, a correction procedure was applied to each im­ ages captured. The images were corrected for back­ ground noise including any variation in the response of the CCD camera pixels. The images were then processed using in house MatLab scripts to extract 2 2.2 Phase Doppler Anemometry Phase Doppler Anemometry (PDA) isa non-intrusive fluid measurement technique used to characterize the size and velocity of distinguish spherical parti­ cles. This technique ismostly used in characterizing aerosols or other types ofjets. PDA is a point mea­ surement, which is not ideal for some experiments. However, the system is coupled with a 2D travers­ ing support system which moves the transmitter and receiver allowing multiple measurement points to be interrogated without realigning the optics. The PDA systems used in this study is provided by Dantec Dynamics. Transmitting optics are composed of a Spectra-Physics Stabilite 2017 laser, which produces a multiline beam with a maximum power output of roughly 7 W . The beam then enters the transmis­ sion box where a quartz prism splits the incoming beam into two 514.5 n m and two 488.0 n m lines. Both lines are connected to a FiberFlow 60 m m 2D transmitter using fiber optics. The receiving optics utilizes a Hidense FiberPDA 57x50 receiver. A BSA P60 Flow and Particle Processor is coupled to the FiberPDA Box. The FiberPDA box is connected to the receiving optics using optical fiber. The Fiber­ PDA box contains 3 photomultiplier tube that con­ verts the optical Doppler burst into current. The current isthen send to the BSA P60 processor where the frequency content of the signal is analyzed by Fast Fourier Transforms (FFT). The processor sends result filesto the post-processing software on a desk­ top computer. (a) Cone angle from sidelight 10 mm Results 3 Q =64.4% 3.1 Qualitative characteristics The spray angle was derived from spray images using (b) Spray solidityfrom LSI side light illumination. The camera exposure time was increased in order to capture the overall droplet Figure 3: Spray characteristics derived from script track. An in-house scriptfilecomputes the cone angle file using the following algorithm: 1. The spray angle ofeach frames isfound by scan­ ning each pixel row of the image from left to right inorder to identifythe pixel locationwhere the light intensity is larger than a user defined threshold. This pixel is then at the left edge of the spray for this row. The same procedure identifies the right edge ofthe spray forthis row. This algorithm islooped through each row ofthe image defined by the user. Figure 3(a) is an ex­ ample of the cone angle derived from the script file. 2 .A In addition to the cone angle,the spray soliditywas determined using LSI. Solidity isdefined by the ratio ofdroplets to empty space within the hollow cone as illustrated in Figure 3(b). The solidityisfound using a script file similar to the one presented above. The advantages ofLSI isthat only droplets located on the light sheet are imaged. Therefore, setting the sheet parallelto the nozzle axis allows an in-depth analysis ofthe hollow cone spray characteristics. Spray solid­ ity has been previously studied in 2006 by Karnawat et. al. [4]. linear regression is fitted to both the left and Table 1 isa summary ofthe qualitative spray char­ right detected spray edges and outputs the cone acteristics. The cone angle was found to increase angle ofthe spray including the deviation ofthe with the increase of water flow rate while varying spray from the nozzle axis. the air flow was found to have minimal impact on the spray geometry. The spray solidity also suggests 3 Table 1: Qualitative characteristic summary Water gph Air Ipm 10 40 20 40 30 40 Cone Angle Solidity% 51.2o 45 58.5o 64 61.7o 77 CFD packages requiresinformation about the surface wave (A) of the primary and secondary break-up re­ gion as well as child droplet diameters. All these pa­ rameters can be found from LSI measurements which are illustrated in Figure 5. Dev. 5.5o 7.4o 8.9o that the hollow cone density increases with the water flow rate. This isexpected asmore mass isdischarged from the atomizer. The side light spray images were captured at a rate of 30 fps. The cone angle was found foreach frame. The derivedinstantaneous cone angle isplotted as a function oftime inFigure 4. The angle fluctuation suggests an RMS value of 7.2o. (a) l g Sa 50 e 40 o ° 30 20 0.5 1 1.5 Time (s) 2 2.5 Figure 4: Cone angle variation as a function oftime The microscopic spray structure illustrated in Fig­ ure 5 is an example of the capabilities of back lit images using LSI. This image shows the largestructure sheets of water formed at the nozzle exit. These sheets break up and form successively smaller droplets as they move away from the nozzle exit. The break-up promoted by the growth and instability of the surface waves can clearly be identified in this im­ age. There are several models that enable the predic­ Figure 5: Microscopic spray characteristics tion of the droplet formation. The structures ob­ served in this image give important information for modelling the spray development. For example, the The macroscopic spray structure illustratedinFig­ K H R T break-up model which is available in most ure 6 demonstrates the different breakup regions in 4 the spray. These images were captured using a laser sheet generated by a rotating prism. The laser beam swept through the field of view at a rate of 48,000 sweeps per minute. This method captures more de­ tails ofthe macroscopic spray structure compared to Figure 3(b). This enables the determination of ad­ ditional model parameters that can be used in spray development modelling. Additionally, these LSI im­ ages can be used to determine the proper Phase Doppler Anemometry (PDA) measurement location downstream of the nozzle exit. This is useful since taking PDA measurements at locations closer to the nozzle would results in errors. The LSI images sug­ gests that large droplet structures exist as far away as 25-26 m m from the nozzle. As such, for this noz­ zle, the PDA measurements are only valid after 10 hydraulic diameters ( 10 D H = 26 mm) away from the nozzle. Only the LSI technique can be used to investigate droplet diameters within the 10 hydraulic diameter region. As willbe seen below, this limit was verified using the PDA technique. 3.2 (a) Instantaneous image from LSI (rotatingmirror) Quantitative characteristics The droplet size statistics and velocity were found using PDA. In addition to the droplet size, the mass flux 1 was derived knowing the volumetric mean di­ ameter (D30) of droplets and the transit time (At) through the measurement volume using Eq. 1 [2]. A polar measurement grid was selected in order to in­ vestigate the cross section ofthe spray (Figure 7). In total, 163 measurement points were interrogated per cross section. Each point recorded a total of 20,000 droplets for proper statistics. N (b) Processedbackground forbetterspraystructure visualization Figure 6: Macroscopic spray characteristics Measurement points Test1 *« X*\ »** 20 *** (1) i 1 The first PDA measurements performed (Test 1) were located on the nozzle axis cross section as illus­ trated by the circled points in Figure 7. This line corresponds to the plane ofthe LSI images discussed 1Volumetricfluxthrough the measurement pointmeasured by thesystem can be convertedtomass fluxknowingtheden­ sityofwater 5 20 m m ' 30mm 60mm Figure 7: Measurement grid above. The PDA measurements were completed in an un-confined atmospheric ambient pressure flow field. The water flow rate was set at 20 gph and the airat a rate of40 Ipm. The selected measurement plane was 10 D h from the nozzle tip. This locationwas selected based on the secondary break-up length identifiedus­ ing LSI as discussed above. The AMD, SMD and flux were analyzed as a function of radial position. Table 2: PDA results summary .z 9 mm 16mm 27 mm Umax D30min 35.2m/s 29.0fim 30.5 m/s 21.4pm 25.5 m/s 18.5pm Derivedflowrate 6.6 13.6 21.4 tipwas found to be largelyasymmetrical with respect to the nozzle axis. The cause ofthis behaviour isstill under investigation. The flux distribution near the nozzle tip appears to be a result of the impingement plate illustrated in Figure 1. The orientation of the impingement plate was alteredto seeifthe locationof the three delivery sections, identified in Figure 9(a), with respect to the swirl air inlet affects the distri­ bution of the mass delivery. However, similar flux distributions were found after this modification. A check of the PDA measurements was performed Figure 8: Spray characteristics of Test 1 at approx­ by integratingthe fluxoverthe complete crosssection imately 10 D h (27 m m ) from the nozzle tip. at successive planes away from the nozzle. The total volume flow rate measured by the flow meter was 20.8 £I!s J- . The totalvolume flowratederived from the PDA measurement integration located at successive The diameter statisticsofthe spray are represented c r o s s s e c t i o n s from the nozzle tip is shown in Table in Figure 8. The lack of flux on the nozzle axis 2 . As can be s e e n , it is only after 27 m m that the demonstrates the hollow characteristic of the spray. i n tegr ate d f l u x i s c l o se to the metered value. This In addition, the flux islargely different from one side suggest that the diameter statistics at distances of ofthe hollow cone to the other resulting in asymmet­ 10 D h and f a r t h e r away from the nozzle tip are valid. rical mass delivery from the nozzle. The maximum flux peak to peak distance from both sides ofthe hol­ low cone was found to be 31 m m . This results in a The contour plots of different spray characteris­ cone angle of59.7o which isidentical to results found tics are demonstrated in Figure 9. These measure­ from LSI. The spray A M D is minimum within the ments were made in a plane at an axial location of hollow cone spray periphery. This isthe result of air 27 m m from the nozzle tip. The most important in­ entrainment from the surroundings to the center of formation extracted from these contours is how the the cone. Droplets having a small Stokes number are differences in droplet characteristics affect the water entrained towards the spray axis leading to smaller mass delivery. From Figure 9(a), the location of the droplet diameter statistics. The minimum A M D and three main flow sections can be identified and cor­ SMD were found to be 14.2 ^ m and 49.4 ^ m respec­ responded to the openings of the impingement plate tively and are located on the spray axis. (Fig. 1). When the spatial position of these sec­ tions iscompared to the data rate, volumetric mean The PDA measurements were conducted in planes diameter and the velocity contour the following con­ 9 to 27 m m distance from the nozzle tip. The flux clusion can be drawn: The data rate is maximum distribution at successiveplanes away from the nozzle (~ 12,000HZ ) at the location of highest flux where 6 Flux cm3/cm2s u = 13.2 m/s, D 3o = 83 ^m. This suggests that the mass is delivered in significantly different ways. For section 1, a small amount of large droplets with low velocity contributed to the flux while for section 2 , a large amount ofsmall droplets with large velocities contributes to the flux. Table 3 is a summary of the characteristic contour analysis. Table 3: Summary of contour plot analysis -30 -20 -10 0 10 20 30 (a) Flux contour D30 i^m 30| -- 1 -- 1 -- 1 -- 1 -- 1 -- Section 1 2 3 4 Conclusion The Delavan swirl-air atomizing nozzles qualita­ tive macroscopic, microscopic and quantitative spray characteristics have been successfully investigated under un-confined atmospheric conditions. Visual­ izations ofprimary and overall spray structures were demonstrated using different imaging techniques. All experiments were conducted at the CanmetENERGY laboratory, Ottawa, Canada. The main conclusions obtained from this study are presented in the follow­ ing points: -3031 ---21 ---11 -- 0 1 --11 --21 --30 1 0 0 0 0 0 (b) Volume mean diameter contour Velocity m/s -3-031 ---21 ---11 -- 0 1 --11 --21 --31 0 0 0 0 0 0 (c) Velocitycontour Flux em2 D30 pm u(m/s) 1.9o" " 74 16~4 3.09 74 20.2 2.52 83 13.2 1. The spray angle for different flow conditions (air to water flow rates) were measured and pre­ sented. The results show that, for the same air flow rate the spray angle increased as the wa­ ter flow rate increased. However for the same water flow rate the spray angle does not change significantly with the air flow rate. 10 2. The cone angle determination from the LSI mea­ surement suggests similar angles with respect to documentation provided by the nozzles supplier. Figure 9: Characteristic contours at z = 27 m m the average velocityis20.2 m/s and the volume mean diameter isD 30 = 74 ^m. The data rate atthe lowest flux section isroughly ~ 8,000 where the average ve­ locity and volume mean diameter are quite different 7 3. Surface waves of the primary ligaments can be recognized and the wavelength of these waves can be estimated or measured from the instan­ taneous LSI images captured using the rotat­ ing prism. These surface waves are responsible for the break-up of liquid columns to liquid lig­ [4] J. Karnawat and A. Kushari. Controlled atom­ ization using a twin-fluid swirl atomizer. Experi­ aments followed by the secondary break-up to ments in Fluids,2006. droplets. 4. The primary and secondary break-up length was [5] D. Kim, O. Desjardins, M. Herrmann, and P. Moin. Towards two-phase simulation of the found to be roughly 5 D H and 10 D H respec­ primary breakup of a round liquid jet by a coax­ tivelyat a flowrateof20 gph ofwater and 40 Ipm i a l f l o w o f gas. Center for Turbulence Research of air. These data are useful in the modelling of Annual research briefs,2006. the break up ofthejetnear the nozzle tip (within the 10 D h). 5. The hollow cone angle derived from the PDA maximum flux location isin agreement with LSI visualization. 6 . The total volume flow rate derived from the in­ tegration of the flux surface at z = 27 m m is in agreement with the water volume flow rate mea­ sured upstream of the nozzle using a rotameter. 7. All spray characteristics demonstrate an asym­ metrical behavior with respect to the nozzle axis. More analysis isrequired to determine the cause ofthis problem. 8 . The PDA spray analysis also demonstrates a large asymmetrical flux distribution for the noz­ zle interrogated. This is believed to have a large impact on the performance ofthese nozzles within gasifiers/burners as the local equivalence ratios are governed by the mass flux distribution of the spray. References [1] E.J. Anthony. Gasification research. CanmetENE R G Y Publications, 2007. [2] Dantec Dynamics. Bsa flow software installation and user's guide v.4.10, 2006. [3] X. Jiang, G.A. Siamas, K. Jagus, and T.G. Karayiannis. Physical modelling and advanced simulations of gas - liquid two phase jet flows in atomization and sprays. Progress in Energy and Combustion Science, 2010. 8