Injector Study via VOF: Emphasis on Vapor Condensation due to Spray As presented at: ILASS–Americas, 23rd Annual Conference on Liquid Atomization and Spray Systems, Ventura, CA, May 2011 W. Kalata*, K. J. Brown, and R. J. Schick Spray Analysis and Research Services Spraying Systems Co. Wheaton, IL 60187 USA Abstract Condensation and absorption are used in gas scrubbing applications and industrial scale production of chemicals. Various processes that depend on condensation techniques require the usage of liquid injection such as jets and sprays. The mass transfer efficiency of these techniques depends on ability of liquid to disperse into vapor and interact in a controlled manner. Computation Fluid Dynamics (CFD) can be used to aid or verify a design of certain class of spray injectors within condensing processes. Volume of Fluid (VOF) technique was used to investigate condensing steam due to cooling by atomized liquid. Two types of hollow cone injectors varying by design and size were investigated at moderate delivery pressure conditions. Initially, both injectors’ spray angles were validated empirically at standard conditions. In this study, the VOF model approaches condensation and evaporation via the Hertz-Knudsen relation. The CFD results revealed a trend that the spray angle decreases as the rate of condensation increases. The increase of the condensation rate was partially dependent on the inverse of characteristic diameter which further was accounted in the VOF mass transfer model. *Corresponding author Experts in Spray Technology Spray Nozzles Spray Control Spray Analysis Spray Fabrication Injector Study via VOF: Emphasis on Vapor Condensation due to Spray Introduction Mass transfer of dispersed liquid has proven to be extremely useful in many industrial applications starting with evaporative flue gas conditioning inside high temperature flow conduits ending with condensing and absorptive systems for production of chemicals. In today’s world, many industrial systems are designed or upgraded to optimize production efficiency with beneficial operating costs. In mass transfer, spray technology provides solutions to many complex problems. more environmentally friendly. While experiments usually can only provide focused point information reliably, CFD can give detailed information of the full computational domain, especially in environments that are difficult to reproduce experimentally. The focus of this study was to investigate spray condensation properties of two differently sized hollow cone nozzles (shown in Figure 1) spraying liquid into condensing vapor. Only a single fluid in its liquid and vapor states were investigated. Presently, Computational Fluid Dynamics (CFD) has been widely adopted in many disciplines that involve critical assessment of fluids in motion. As technological advancements in numerical simulation have developed, fluid motion is often coupled with complex heat and mass transfer. Currently, simulation tools are becoming more available and cost effective in engineering applications. The computational sciences have become an integral part of engineering research and design. Figure 1. Spraying Systems Co. pressure swirl type spray nozzles (small capacity – left, large capacity – right). Compared with empirical testing, numerical simulation is becoming more cost effective and METHODS Swirling Nozzle Simulations The CFD simulations were performed with ANSYS FLUENT version 12.1. The CFD models were reproduced according to the internal geometry inside swirling chambers and orifices for two injectors that are shown in Figure 1. The biggest difference was the spray nozzle capacity. The orifice of the large capacity injector was two orders of magnitude larger than the small capacity nozzle. In some instances, geometry was simplified to reduce geometrical complexities in 3D CAD model (Figure 2). Figure 2. Swirling injector chambers used in CFD. Meshing was performed within ANSYS Workbench 13. Initially each unstructured grid was composed of about 4.0–4.5 million of mixed cells which employed boundary layer type inflation at all walls. Inside FLUENT, each mesh was converted into polyhedral grid while the boundary layer mesh remained. The grids were reduced to 0.8–1.2 million polyhedral cells. Dense mesh was incorporated in vicinity of orifices. Size functions were used to focus smaller grid sizes on geometries such as whirling chambers or orifices. Each CFD model was set up with liquid velocity or mass flow inlet boundary condition (BC) with temperature at 20ºC. The outlet pressure BC was setup as constant zero pressure with standard 1 bar operating pressure. The backflow temperature for 2 www.sprayconsultants.com Injector Study via VOF: Emphasis on Vapor Condensation due to Spray METHODS the outlets was set as slightly above saturation temperature (Tsat) (105ºC) with only vapor passing through, which means that the liquid volume fraction backflow (LVFBF) was set to zero. Injector and other walls were set as rigid, with no-slip and adiabatic conditions. All vapor material properties were dependent on temperature. The density of the vapor was solved through incompressible ideal gas law, where pressure term is an operating pressure. Throughout all simulations the following models were included: k-e Realizable Turbulence Model, energy enabled and Volume of Fluid (VOF) with inclusion of mass transfer and temperature dependent surface tension of liquid. The simulations were performed in steady state mode. Table 1 shows the BC summary for both injectors used in this study. Nozzle Dorifice Prated* Inlet BC VIN AIN QIN QIN ṁIN Dh Re TIN Outlet BC Type POUT TBF LVFBF TX-2 CRC-250 mm barg 0.71 3.79 82.6 2.07 m/s m2 m3/s lpm kg/s m – °C 5.24 4.7054E-07 2.4656E-06 0.1479 0.002461 0.0007740 3.30E+04 20 6.48 0.004927 0.03193 1915.7 31.871 0.07921 4.18E+06 20 bara °C barg Mass Transfer L kJ/kg Tsat °C βqp** Dq (SMD)* μm B - The VOF model used implicit scheme for volume fraction in equation below (1) with modified version of High Resolution Interface Capturing (HRIC) discretization scheme [1]. (1) 𝛥𝛥𝛥𝛥 + 𝑓𝑓 = 𝑆𝑆𝛼𝛼 + 𝑛𝑛+1 𝜌𝜌𝑞𝑞𝑛𝑛+1 𝑈𝑈𝑓𝑓𝑛𝑛+1 𝛼𝛼𝑞𝑞,𝑓𝑓 = 𝑛𝑛 𝑝𝑝=1 𝑚𝑚𝑝𝑝𝑝𝑝 − 𝑚𝑚𝑞𝑞𝑞𝑞 In VOF scheme equation (1) above, ṁqp is the mass transfer from phase q to phase p and ṁpq is the mass transfer from phase p to phase q. By default, the mass source term on the right-hand side of Equation (1) Sα, is zero, but you can specify a constant or user-defined mass source for each phase. Uf, ρq, ρp, αq,f , Δt, V, f, and n are volume flux through face, density of phase q and p fluid, face value of phase volume fraction, time step (or iteration step if steady), volume of cell, cell face index, and computational stepping index [1]. In this study, right side from equation (1) was employed to assess mainly condensation effects within the vicinity of sprayed liquid phase. The mass transfer was conditional with respect to the saturation temperature (Tsat) that was an input by a user. In most cases, for water liquid and vapor, Tsat was set at 100°C since operating pressure was set at standard conditions. In higher pressure applications, Tsat would have to be determined based operating pressure for particular fluid that is being analyzed. Additionally, volumetric heat source (Sh) was included in energy computations. It was computed based on calculated mass transfer for a particular phase (ṁpq) and fluid’s latent heat (L) as shown in equation (2) below. Constant Pressure 1 1 105 150 0 0 2257 100 0.01 1620 48.2 𝛼𝛼𝑞𝑞𝑛𝑛+1 𝜌𝜌𝑞𝑞𝑛𝑛+1 − 𝛼𝛼𝑞𝑞𝑛𝑛 𝜌𝜌𝑞𝑞𝑛𝑛 (2) 2270 100 0.01 78 1002 ̇ *laboratory measurement **Hagen et al. [7] Table 1. Boundary conditions summary 3 www.sprayconsultants.com Injector Study via VOF: Emphasis on Vapor Condensation due to Spray METHODS Mass Transfer Considerations (9) The mass transfer model (MTM) used in VOF simulations was based on supplying evaporation and condensation constants (A and B respectively) in mass transfer equations shown in (3) and (4) [1]. These equations represent mass rate per volume (kg/s•m3) transferred through gas-liquid interface. When setting p to be gas phase and q liquid phase, ṁqp becomes mass rate from liquid to vapor (evaporation), ṁpq represents mass rate from gas to liquid (condensation), αp and αq are volume fractions for gas and liquid, and ρp and ρq are densities for gas and liquid. (3) (4) ̇ ( ( (8) (7) ) ( ⁄ ) ( )( )√ ⁄ ) (11) ( ( ) (10) β, M, R, Tsat, Psat, and P* are accommodation coefficient, molecular weight, universal gas constant, saturation temperature and pressure, and partial pressure at the interface, respectively. When assuming that pressure and temperature (P* and T *) are close to saturation condition, a combination of (5) with Clasius Clapeyron relation (6) and accommodation for interfacial area within cell volume (7), resulted with derivation for equations (3) or (4), where A or B coefficients take form of equations (8) and (9). (6) ( In various scientific literatures, theoretical and experimental values for evaporation or condensation accommodation coefficients (βqp and βpq respectively) can be found. There, accommodation coefficients are further modified to equations (10) and (11) shown below where Ke and Kc are known as evaporation and condensation constants [4-7]. As ANSYS FLUENT documentation points out, A and B need to be fine-tuned accordingly to fluids’ properties while Hertz-Knudsen relation (5) is considered. This formula gives evaporation-condensation flux (F) based on kinetic theory [2-5]. √ )√ L, A i, Dp and Dq are latent heat, interfacial area density, volume fraction of fluid, and representative diameters for both condensation and evaporation, respectively. By quick examination, it can be noticed that density term can make A larger than B by 2 or 3 orders of magnitude, depending on fluid and operating conditions which may vary saturation temperature. Also Dp and Dq could be assumed based on SauterMean Diameter (SMD, D32) measured at that specific flow rate at standard laboratory conditions. As it can be noticed in equations (8) and (9), A and B are proportional to an inverse of SMD. ̇ (5) ( )( In studies with whirling injectors, water and steam were used to determine injector’s performance when condensation effects are involved. Condensation and evaporation constants for water were available in literature where these constants were obtained through various experimentally and theoretically [4-7]. The value for condensation accommodation coefficient was 0.01 which was chosen based on Hagen et al. [7]. The simulations were performed first without and then with MTM to capture effects of condensation with respect to non-condensing environment. A comparative study was performed indicating spray performance differences. ) ) 4 www.sprayconsultants.com Injector Study via VOF: Emphasis on Vapor Condensation due to Spray METHODS Backflow Study Large capacity pressure swirl type of nozzle (see Figures 1 and 2) was used in studying an injector outflow in various liquid volume fraction backflow conditions ranging from 0.0 to 0.05. This backflow was a part of constant pressure outlet BC setup. Overall schematic of the flow problem is shown in Figure 3. In this study, the level of liquid volume fraction (LVF) that advances into swirling core inside the nozzle chamber was varied until the spray collapse. The advancement is due to negative pressure caused by swirling fluid inside and below the nozzle. A toxic fluid mainly composed of trioxin (Metaformaldehye) ~50%, water ~25% and additional toxic chemicals was used in cases with inlet gas velocity BC of 4 and 8 m/s. Mass exchange model described above was enabled. Figure 3. CFD setup schematic for LVF backflow study. RESULTS & DISCUSSION angle resulted in ~90 degrees for both cases with and without MTM. Swirling Nozzle Simulations The summary of results for low capacity injector simulation with mass transfer is shown in Figure 5 below. The results show partial collapse of the spray plume as compared to the simulation performed without MTM (see Figure 6). Initially, in the case without condensation, the simulation resulted with ~66 degrees. When condensation was included, the temperature of vapor advancing up from the bottom towards the orifice had a temperature drop over a longer span, therefore cooling it earlier (Figure 6 top). The spray angle resulted in ~47 degrees (Figure 7 bottom). Backflow Study At gas velocity approaching a spray at 4 m/s, the spray collapsed at backflow LVF value of 0.05 (Figure 9 top). With gas velocity at 8 m/s, the spray collapsed partially at LVFBF value of 0.04 (Figure 9 bottom). Discussion In low capacity injector study, an inlet BC and injector geometry was set up according to work by Bade et. al. The difference between the reported study by Bade et al. and current study without condensation was the gaseous phase which was water vapor instead of air at laboratory conditions and thermal effects were added with vapor properties varying with temperature. The case without condensation showed some differences in dynamics as compared to Bade et al. work. The spray angle was 66 degrees as compared to 77 degrees reported by Bade et al. Taking account that injector is spraying into less dense gas and with higher viscosity, it experienced The summary of results for high capacity injector simulation with MTM is shown in Figure 7 below. The results show no collapse of the spray plume with respect to the simulation performed without mass exchange (see Figure 8). Slight variations in temperature were noticed in the swirling core inside the nozzle chamber and in the bottom portion of the spray. Temperature profiles of fluid mixture were slightly higher with MTM where condensing process added heat to the mixture via latent heat. The spray 5 www.sprayconsultants.com intuitive scenario where nozzle performance depends on both heat and mass transfer which are directly coupled. In the large capacity injector case, interaction dynamics are significantly different. There is a significant swirling core inside the nozzle chamber and its size has remained the same when comparing cases with or without MTM. There were slight temperature effects due to condensation heat sources but the overall dynamics remained almost identical. The additional study showed that liquid volume fraction would have to be higher than 0.04 or 0.05 inside the core to decrease the spray angle or even colcases withspray or without There were slight lapse the plume.MTM. For this to occur, there needs to temperature effects due to condensation be either a high condensation rate inside theheat core, which sources the overall almost is rather but unlikely due todynamics the natureremained of the spray, or high enough LVF advanced due to suction of the swirling identical. action. The additional studyforshowed that liquid volume There is room improvement in this model. The fraction be higher than 0.04 ordiameter 0.05 as MTM atwould some have point to relies on characteristic inside core to decrease the. spray evenwas shownthe in equations (8) and (9) In thisangle studyor SMD used for the the spray characteristic diameter. It wasthere based on collapse plume. For this to occur, laboratory the spray at rate the inside same flow needs to bemeasurement either a high of condensation the rate but in standard conditions. TheofSMD core, which is rather laboratory unlikely due to the nature the changes and throughout the temporal evolution spray, or spatially high enough LVF advanced due to suction of of the spray. Also droplet growth is not accounted for the swirling action. in MTM. It is likely that mass and heat sources will decrease as droplets experience growth. There is room for improvement in this model. The Discussion In low capacity injector study, an inlet BC and injector geometry was set up according to work by Bade et. al. The difference between the reported study by Bade et al. and current study without condensation was the gaseous phase which was water vapor instead of air at laboratory conditions and thermal effects were added with vapor properties varying with temperature. The case without condensation showed some differences in dynamics as compared to Bade et al. work. The spray RESULTS & DISCUSSION angle was 66 degrees as compared to 77 degrees reported by Bade et al. Taking account that injector is spraying into less dense gas and with higher viscosity, it a resistance awhich posedwhich a result that ais result somewhat experienced resistance posed that is counter intuitive (density and dynamic viscosity somewhat counter intuitive (density and dynamic vis3 and 1.25e-5 N·s/m of steam 105°Catis105°C 0.58 kg/m 1.25e-5 cosity of atsteam is 30.58 kg/m3 and 3 respectively and air and density and at viscosity N·s/m respectively and air density viscosity 20°C is at 3 3 3 and N·s/m 1.79e-5 N·s/m3 respective20°C is and and 1.21 kg/m1.21 andkg/m 1.79e-5 respectively). ly). A quick studystudy was was performed where steam A quick performed where steam tempertemperature was increased consequently decreasing ature was increased consequently decreasing the steam density butdensity increasing dynamicthe viscosity and further the steam but the increasing dynamic causing of spray angle. The plot below angle. (Figviscosityreduction and further causing reduction of spray ure shows that the spray anglethat decrease and angle corresThe4) plot below (Figure 4) shows the spray ponding kinematic vapor/air viscosity kinematic increase with decreasevapor/air and corresponding increasing temperature. viscosity increase with increasing temperature. Injector Study via VOF: Emphasis on Vapor Condensation due to Spray MTM at some point relies on characteristic diameter Conclusions as shown in equations (8) and (9). In this study SMD In this the performance of pressure was used forstudy, the characteristic diameter. It was swirl atomizers in condensing fluid were investigated. Two based on laboratory measurement of the spray at the hollow cone spray nozzles with significant variations in same flowcapacity) rate but inwere standard laboratory conditions. size (and simulated and compared with The SMD changes spatially and throughout the non-condensing scenarios. temporal evolution of thenozzle spray.was Also dropletbygrowth The small capacity affected the conisdensation not accounted MTM. angle It is likely that mass and where for itsinspray partially collapsed heat sources will decrease as droplets (Figures 6 and 10). The large capacityexperience spray was not affected directly by condensation (Figures 8 and 10). growth. However when liquid volume fraction was introduced into swirling core inside the nozzle chamber, the spray Conclusions Figure Spray angleangle and kinematic viscosity viscosity with varyingwith Figure4. 4. Spray and kinematic temperature. varying temperature. plume was affected and even collapsed at LVF of 0.05 In this study, (Figure 9). the performance of pressure swirl The mass transfer model reliedinvestigated. on fluid properties atomizers in condensing fluid were Two and on the characteristic diameter that was based hollow cone spray nozzles with significant variationson The coefficient for the condensation was inSMD size values. (and capacity) were simulated and compared proportional to the inverse of the characteristic diamewith non-condensing scenarios. MTMMTM was was not included. Steam density was was calculated not included. Steam density calcuwith ideal gas law shown in equation (12). lated with ideal gas law shown in equation (12). (12) ter; therefore theoretically the smaller the drop size, the (12) better the capacity condensation rate. The small nozzle was affected by the condensation where its spray angle partially Nomenclature collapsed (Figures 6 and 10). The large capacity spray was not affected directly by condensation (Figures face index 8f and 10).cell However when liquid volume fraction n computational stepping index was introduced into swirling core inside the nozzle inlet mass transfer flow rate chamber, the spray plume affected and peven mass transfer fromwas phase q to phase collapsed at LVF of 0.05 (Figure 9). As condensation waswas included thethe spray angle As condensation included spray angle colcollapsed even further as shown in Figure 5. When lapsed even further as shown in Figure 5. When kinekinematic viscosity of mixture the mixture investigated matic viscosity of the was was investigated around around the local orifice, local kinematic wasdue lower the orifice, kinematic viscosityviscosity was lower to condensed vapor. vapor. This isThis a reversed situation as comdue to condensed is a reversed situation pared to non-condensing scenario. It poses nonas compared to non-condensing scenario. It poses nonintuitive scenario where nozzle performance depends on both heat and mass transfer which are directly coupled. The mass transfer model relied on fluid properties and on the characteristic diameter that was based on SMD values. The coefficient for the condensation was proportional to the inverse of the characteristic diameter; therefore theoretically the smaller the drop size, the better the condensation rate. In the large capacity injector case, interaction dynamics are significantly different. There is a significant swirling core inside the nozzle chamber and its size has remained the same when comparing 6 www.sprayconsultants.com Injector Study via VOF: Emphasis on Vapor Condensation due to Spray Nomenclature Prated Psat Q IN R Re pressure rated at certain flow rate fluid saturation pressure inlet volumetric flow rate universal gas constant Reynolds number S α mass source for energy equation S h volumetric heat source for energy equation T temperature T* partial temperature at the phase interface T BF backflow temperature T IN inlet temperature T sat fluid saturation temperature U f volume flux through face V volume of cell V IN velocity for inlet BC f n cell face index computational stepping index ṁIN inlet mass transfer flow rate ṁqp mass transfer from phase q to phase p ṁpq mass transfer from phase p to phase q p first phase (gas) q second phase (liquid) A A i A IN B D32 D h Dorifice D p D q F K c K e L LVF BF M P* POUT evaporation constant interfacial area density inlet BC cross-sectional area condensation constant Sauter-Mean Diameter hydraulic diameter orifice diameter representative diameter in phase p representative diameter in phase q evaporation-condensation flux condensation constant evaporation constant fluid latent heat of vaporization backflow liquid volume fraction molecular weight partial pressure at the phase interface outlet pressure αq,f αp αq β βpq βqp Δt ρp ρq face value of phase volume fraction volume fraction of phase p volume fraction of phase q accommodation coefficient accommodation coefficient for condensation accommodation coefficient for evaporation time step (or iteration step if steady) density of phase p density of phase q References 1. ANSYS FLUENT 12.0 - Theory Guide, ANSYS, Inc., Canonsburg, PA, 2009. 2. Ythehus, T. and Ostmo, S., International Journal of Multiphase Flow, 22-1:133-155 (1996). 3. Wang, Z.-J., Chen, M. and Guo, Z.-Y., International Conference “Passive and LowEnergy Cooling for the Built Environment:, Santorini, Greece, May 2005, pp. 543-547. 4. Tamir, A. and Hasson, D., The Chemical Engineering Journal , 2: 200-211 (1971). 5. Marek, R and Straub, J., International Journal of Heat and Mass Transfer, 44: 39-53 (2001). 6. Mills, A.F. and Seban, R.A., International Journal of Heat and Mass Transfer, 10: 1815-1827 (1967). 7. Hagen, D.E., Schmitt, J., Trueblood, M., Carstens, J., White, D.R. and Alofs, D.J., Journal of the Atmospheric Sciences, 46-6:803-816 (1989). 8. Bade, K.M., Kalata, W. and Schick, R.J., ILASS Americas, 22nd Annual Conference on Liquid Atomization and Spray Systems, Cincinnati, OH, May 2010. 7 www.sprayconsultants.com Injector Study via VOF: Emphasis on Vapor Condensation due to Spray Figure 5. Summary of CFD results for the small capacity injector. 8 www.sprayconsultants.com Injector Study via VOF: Emphasis on Vapor Condensation due to Spray Figure 6. Main differences between cases with and without mass transfer in the small capacity injector. 9 www.sprayconsultants.com Injector Study via VOF: Emphasis on Vapor Condensation due to Spray Figure 7. Summary of CFD results for a large capacity injector. 10 www.sprayconsultants.com Injector Study via VOF: Emphasis on Vapor Condensation due to Spray Figure 8. Noticeable differences between cases with and without mass transfer in a large capacity injector. Figure 9. VOF results for backflow study. Effect of LVF backflow on the large whirling nozzle with two gas velocity modes. 11 www.sprayconsultants.com Injector Study via VOF: Emphasis on Vapor Condensation due to Spray Figure 10. Streamlines indicating swirling liquid motion inside and outside the nozzles. 12 www.sprayconsultants.com
© Copyright 2026 Paperzz