3D-CFD lagrangian spray simulations for large two stroke marine diesel engines compared with experimental data of aspray combustion chamber

Proceedings of the ASME 2012 Internal Combustion Engine Division Spring Technical Conference ICES2012 May 6-9, 2012, Torino, Piemonte, Italy ICES2012-81016 3D-CFD LAGRANGIAN SPRAY SIMULATIONS FOR LARGE TWO STROKE MARINE DIESEL ENGINES COMPARED WITH EXPERIMENTAL DATA OF A SPRAY COMBUSTION CHAMBER M. Bolla ETH Zurich, Zurich, Switzerland Y. M. Wright ETH Zurich, Zurich, Switzerland M. A. Cattin ETH Zurich, Zurich, Switzerland K. Boulouchos ETH Zurich, Zurich, Switzerland ABSTRACT The behavior of lagrangian spray models for the application in large two stroke marine engines is investigated. 3D-CFD simulations of a Spray Combustion Chamber (SCC) with a single hole (0.875 mm diameter) injector are presented and compared with experimental results. Shadow images of the spray under evaporating and non-evaporating conditions, with and without swirl flow and for different chamber pressures are available by means of which the simulation results are validated. A novel post processing methodology for 3D CFD spray simulations is introduced, which converts the numerical data into images which allows for a more rigorous quantitative comparison with the experimental data. Good agreement of the simulation results with the experiment is reported both in terms of spray penetration as well as concerning the evaporation of the fuel. Since the appropriate discretization of the large volumes typical of 2-stroke marine engines presents a substantial challenge, the influence of the grid resolution is investigated. In addition, the influence of fuel quality on the evolution of the spray morphology is assessed. For this purpose, simulations with heavy fuel oil (HFO) are compared with experiment. INTRODUCTION Large Diesel engines are commonly employed for merchant ships propulsion due to the high thermal efficiency and the relative low fuel costs for HFO. Since 2000, exhaust gas emissions from maritime Diesel engines are legislated [1]. As a consequence, much research activity is currently underway to substantially reduce nitric oxides from those combustion R. Schulz Wärtsilä Switzerland Ltd, Winterthur, Switzerland systems. Pollutant formation is strongly influenced by the state of mixing during combustion. The fuel spray characteristics play an important role for mixture formation and the subsequent combustion since they determine droplet sizes and positions and therefore influence the evaporation and fuel/air mixture processes. The spray itself depends strongly on the fuel properties, which can vary widely between light fuel oils such as Diesel and low quality HFO. The fast increase in computation power availability in recent years has promoted the use of Computational Fluid Dynamics (CFD) as a powerful tool in assisting the design of advanced combustion systems. In the last decades, various model approaches for two-phase flows description have been proposed. For high-pressure dispersed sprays typical of Diesel engines, the Eulerian-Lagrangian methodology for gas- and liquid-phase, respectively is widely used. In this approach – which is also adopted in this study – the conservation equations are solved for both phases and are coupled by introducing source terms which account for the exchange of momentum, energy and mass. Roughly two thirds of merchant ships are operated with heavy or residual fuel oil, for which the composition and hence thermo-physical properties can vary substantially [2]. Most experimental investigations for low grade fuels reported in the literature have been carried out at the visual constant volume combustion chamber (VCVCC) [3] and at the fuel ignition analyzer (FIA) [4]. The former has a cylindrical shape with 150 mm bore and 350 mm in height with optical access to the full length of the spray, while the FIA has a bore of 65 mm and 1 Copyright © 2012 by ASME is 190 mm long without optical access. Both devices have an axial injector with a hole diameter of 0.16 mm. HFO spray simulations have been performed by different research groups. Goldsworthy [5] developed a simplified model for HFO consisting of residual base and cutter stock. Ignition delay and flame propagation were compared to measurement at the VCVCC presented in [5]. Struckmeier et al. [4, 6] improved the model of Goldsworthy including multi-component evaporation modeling. Good predictions of flame propagation and lift-off height at the VCVCC were reported. In [7] Kyriakides et al. developed a model for HFO spray accounting for different HFO thermo-physical properties. The influence of fuel properties were assessed purely numerically in the absence of experimental validation data. Two different injector hole sizes were employed, namely 0.37 and 0.9 mm, representing orifice diameters typical of large 4-stroke and 2-stroke engines, respectively. HFO showed comparable penetration length compared to light Diesel fuel and significantly larger droplet sizes in agreement with findings reported in [2]. More recently, Chryssakis et al. [8] performed reactive simulations at the VCVCC and FIA by adopting the HFO model proposed in [7]. In the present investigation, non-reactive numerical simulations of light Diesel fuel and HFO spray injection are carried out for a broad range of conditions typical to large 2-stroke marine engines. Results are compared to optical measurements available from a new experimental device representative for large two-stroke marine Diesel engines [9, 10]; a production grade fuel injector equipped with a single coaxial nozzle with an orifice diameter of 0.875 mm was used. The main objective of the present study is to assess if the spray models employed – which have been developed for considerably smaller dimensions (automotive and heavy-duty) – are applicable also at the dimensions and time-scales of large marine engines. TEST CASES Experimental data available from the Spray Combustion Chamber (SCC) installed at Wärtsilä Switzerland Ltd. [9-11] have been used for model validation at conditions relevant to large two-stroke marine Diesel engines. The optically accessible chamber has a disk shape with 500 and 150 mm in diameter and height, respectively. Fuel is injected from the periphery into swirling ambient air or N2. The test facility is able to operate with different fuel qualities from light Diesel fuel up to HFO. For more detailed information about the SCC test facility, the reader is referred to [9, 10]. Nine different cases have been considered (cf. Table 1), which consist of evaporating and non-evaporating conditions, three different ambient gas densities and light Diesel fuel and heavy fuel oil. All test cases are operated with a one-hole coaxial injector nozzle with a hole diameter of 0.875 mm. Injection pressure and duration are 1000 bar and 25 ms, respectively. Table 1: Operating conditions considered. LFO LFO & HFO Gas density [kg/m3] Gas pressure [bar] 11 22 33 11 22 33 13 27 40 30 60 90 Gas temperature [K] 400 900 NUMERICAL SETUP Flow-field solver Simulations have been performed using the commercial CFD-solver Star-CD [12], a finite-volume solver for fully unstructured grids. Conservation equations are solved with the RANS formulation using the RNG variant of the k-e turbulence model with default model constants. The SCC numerical domain is illustrated in Figure 1. It is a closed geometry, which consists of inlet tube (blue), regenerator and tilted flange (red) and disk-shaped combustion chamber (green). Red circles represent the position and size of the window, while the black arrow indicates the location of the injector and direction in which fuel enters the chamber through the single co-axial orifice. Figure 1: Spray Combustion Chamber numerical domain consisting of inlet (blue), regenerator and flange (red) and combustion chamber (green). Arrows indicates location and direction of injector. In the combustion chamber a hexahedral grid is employed. The entire numerical domain consists of 0.97 million cells. The region relevant for the spray, 300 mm axial and 50 mm radial from the spray axis has been refined with a homogeneous grid size of 2 mm, corresponding to roughly twice the injector diameter. Initial field quantities (temperature, pressure, velocity, turbulent kinetic energy and turbulence dissipation rate) at start of injection (SOI) have been computed from a previous simulation, where the experimental filling process of the chamber using pressurized air has been computed [13, 14]. The tilted intake into the combustion chamber provides the desired swirl motion in the combustion chamber, which at SOI corresponds approximately to a solid body rotation with 20 m/s at 200 mm radial position and a no-slip condition at the wall. Heat transfer at the walls is modeled with the standard formulation of wall-function [12]. A constant wall temperature (TWall=453 K) boundary condition is employed. 2 Copyright © 2012 by ASME Spray model Fuel spray is treated with the built-in Eulerian-Lagrangian approach. Primary break-up is modeled according to [15] and secondary break-up with the Reitz-Diwakar [16] model, where two different break-up regimes are considered, i.e. the bag and the stripping break-up regime. Simulations are computed with unchanged model constants as proposed Star-CD default values. Standard models for momentum exchange, droplet collision, evaporation, droplet-wall interaction are used [12]. Liquid fuel properties needed for exchange processes with the gas phase (saturation pressure, density, surface tension, heat capacity, heat of evaporation, viscosity and conductivity) are tabulated in temperature using internal n-dodecane properties for light Diesel fuel and for HFO simulations fuel properties from Wärtsilä Switzerland Ltd. (private communication by Dr. R. Schulz). HFO consists of five different components, which are treated with the Star-CD built-in multi-component spray model approach [12]. POST-PROCESSING METHODOLOGY The simulated spray penetration length is evaluated in an analogous way as in the experiment [9], where using shadow imaging the spray is separated into a dense and a contour region, which are defined as the 10 % and 90 % value, of the transmitted light intensity through the spray, respectively. With this method, additional information from the measurement under evaporating conditions for validation purposes is obtained. In the following, the methodology is shortly explained. The physics of shadows imaging is tried to be reconstructed, mainly light crossing the spray and look what does come through in order to evaluate spray optical thickness. The plane perpendicular to the experimental light beam direction trough the spray is discetized into an equidistant cartesian grid with a resolution of 0.4 mm. In each grid node the sum of the projected droplet surface areas present in the optical path is considered as representative shielding area. Light is assumed not to be transmitted through liquid droplets. The transmitted light ratio, TR , is introduced as: TR = ∑A dropi i (1) Agrid By considering each grid node independently, the effect of light scattering due to droplets is not taken into account. Former phenomenon is considered in an artificial way through a smoothing treatment of the transmitted intensity ratios between the neighboring cells as follows: TR(i , j ) ( = smooth (1 − χ ) TR(i , j ) 1 + χ TR(i , j +1) + TR(i , j −1) + TR(i −1, j ) + TR(i +1, j ) 4 ) (2) where χ is the smoothing rate set arbitrarily to be 0.5. The smoothing treatment does have the physical meaning of light scattering and does not influence the macroscopic structure of the spray. The transmitted light ratio can be now compared directly with the grayscale images taken with the experimental high-speed camera. Identically as in the experiment, the simulated dense and contour spray regions are defined as locations where TR is higher than 90 % and 10 %, respectively. Penetration length of the dense spray is defined as highest distance from the injector where a minimal dense region of thickness of 5 d0 is present as illustrated in Figure 2. Figure 2: Minimal thickness definition for dense spray length. Definition of the contour spray penetration length is analogous to the one of the dense spray. RESULTS AND DISCUSSION Grid sensitivity Numerical simulations of two-phase flows with the Eulerian-Lagrangian framework are well known to be grid dependent [17-20]. However, it is a common praxis to use this method for Diesel spray injection because of its simplicity and low computational costs [21]. The grid sensitivity is evaluated for a reference case with 33 kg/m3 gas density at nonevaporating conditions in order to evaluate the break-up model independently, i.e. in the absence of any evaporation effects. In the region relevant for the spray, the CFD mesh consists of perfectly hexahedral cells with homogeneous size. Three different sizes have been tested, namely 1, 2 and 4 mm side length, leading to 3.6, 0.97 and 0.6 Mio cell counts for the entire domain, respectively. The comparison of the spray penetration lengths is given in Figure 3: With a finer mesh a faster penetration of the spray for the first 2 ms is observed. Droplet momentum is exchanged with smaller cell volumes and the transfer of momentum in the radial direction is considerably reduced compared to the coarser grid. 3 Copyright © 2012 by ASME Figure 3: Penetration length of light Diesel fuel under nonevaporating conditions with 33 kg/m3 gas density for three different grid resolutions: 1 mm (red), 2 mm (green) and 4 mm (red). leading to faster penetration in the first 200 mm. After this distance the trend of spray penetration length is reversed, i.e. longer penetration for the coarser grid, for which a higher gas velocity magnitude in the region farther than 200 mm from the injector is observed. The numerical shadow images shown in Figure 4 (lower) present the influence of the grid resolution on the liquid droplets distribution. With the finest mesh, the velocity field shows more vortices which increase radial convective transport. A substantially broader and more dispersed spray can be observed and consequently the axial penetration is decreased. For the given injection configuration at the investigated operating conditions, the best agreement can be obtained at a grid resolution of 2 mm, which corresponds to roughly twice the injector orifice diameter. This is in agreement with previous studies reporting sensitivities of spray morphology evolution on grid resolution [19, 22]. For this resolution the weakest sensitivity on spray model constants was also observed (not shown here). It can be concluded, that reasonable results can be obtained using a “standard” implementation of the ReitzDiwakar spray model with a 2 d0 grid also for substantially larger injector sizes, despite the fact that the model was not originally developed for such large injector diameters [16]. Non-evaporating conditions Atomization and secondary break-up model performance are evaluated under non-evaporating conditions in order to suppress droplet mass transfer. Note that dense and contour spray are the same under cold ambient gas. Figure 5 shows light Diesel spray penetration length for three different gas pressures. The increase in penetration when reducing the gas density is correctly captured and the good agreement confirms the appropriateness of the mesh resolution and spray model constants. The upper two sprays in Figure 6 compare simulated and measured shadow images under non-evaporating conditions at 5 ms after SOI and 40 bar ambient pressure, respectively. Interesting to note is the lateral deviation due to the swirling ambient gas. Roughly 200 mm downstream, deviation in the opposite direction occurs, since swirl direction is reversed compared to the spray axis. This effect is well reproduced by the simulation. Figure 4: Computed gas phase velocity (colored figure) and computed shadow image (gray figure) of light Diesel fuel under non-evaporating conditions with 33 kg/m3 gas density at 5 ms after SOI for three different grid resolutions: 1 mm (upper), 2 mm (middle) and 4 mm (lower). As a consequence, higher gas velocities are induced in the inner of the spray as can be observed in Figure 4 (upper) 4 Copyright © 2012 by ASME (a) p=30 bar, T=900 K Figure 5: Comparison simulation (dotted lines) with experiment (straight lines) of penetration length with light Diesel fuel under non-evaporating conditions for three different gas pressures: 13 bar (blue), 27 bar (green) and 40 bar (red). Non-evaporating, simulation Non-evaporating, experiment (b) p=60 bar, T=900 K evaporating, simulation evaporating, experiment Figure 6: Shadow images of light Diesel fuel under nonevaporating (upper two) and evaporating (lower two) conditions with 33 kg/m3 gas density and 5 ms after SOI. Simulation at the top and experiment at the bottom. Evaporating conditions Experiments under evaporating conditions are conducted with nitrogen as process gas to ensure a non-reactive environment. The same three gas densities as for the cold cases have been investigated to ensure that the aerodynamic forces acting on the droplets remain unchanged in order to investigate the effect of evaporation. (c) p=90 bar, T=900 K Figure 7: Comparison simulation (blue) with experiment (red) of dense (pointed lines) and contour (straight lines) spray penetration length of light Diesel fuel under evaporating conditions for three different gas pressures: (a) 30 bar, (b) 60 bar and (c) 90 bar. 5 Copyright © 2012 by ASME Penetration lengths of the dense and contour spray for the three different ambient pressures are visualized in Figure 7 for light Diesel. Following an identical evolution at the early stages, a separation between the dense spray from the contour thereof is evident, which is due to rapid evaporation of the small droplets resulting from secondary break-up. The simulation is capable to capture this effect and the trend of an increased separation distance by reducing gas pressure is qualitatively correctly reproduced. Separation lengths are however moderately overpredicted for all gas densities (22, 23 and 19 % for 90, 60 and 30 bar, respectively). Compared to experiments, the simulations show a weaker separation of the dense part from the spray tip contour. It is unclear if these differences stem from the numerical spray solution procedure itself or from the definition of the dense spray length during the post-processing; further investigations of evaporating sprays are required to clarify the origin of the discrepancies. The two sprays shown above in Figure 6 (lower) compare simulated and measured shadow images for the 90 bar case. Droplet evaporation is almost completed after approximately 160 mm and the predicted spray shape qualitatively agrees well with the experiment. Although no measurements concerning fuel vapor distributions are available for comparison, the predicted fuel vapor mass fraction iso-contours at 5 ms after SOI are presented in Figure 8 to illustrate the pressure influence on the evaporation. The structure is conceptually similar to that of sprays from injectors with smaller diameters, with a dense and a dilute spray at 50-100 d0 from the injector [23]. enhanced by reducing injection pressure and by increasing injector orifice diameter [25]. Heavy fuel oil Recent measurements performed with HFO injection at 900 K ambient temperature did not show a separation between dense and contour spray because of the poor evaporation characteristics of the fuel [10]. The fuel is discretized into five components for which the six liquid properties required by the numerical model (surface tension, cp, density, viscosity, latent heat and vapor pressure) are tabulated in temperature. The evaporation is then treated by the built-in multi-component model [12]. HFO penetration depth comparisons for three different ambient pressures are presented in Figure 9. As for the Diesel cases discussed above, the dependence of the gas density is well described, although for increasing pressure, the simulation progressively overestimates the penetrations. Figure 9: Comparison simulation (dotted lines) with experiment (straight lines) of penetration length with HFO at 900 K gas temperature and three different gas pressures: 30 bar (blue), 60 bar (green) and 90 bar (red). Figure 8: Light Diesel fuel vapor mass fraction distribution at 2 ms after SOI and 900 K gas temperature with different ambient pressures: 30 bar (upper), 60 bar (middle) and 90 bar (lower). A fuel rich region is formed at the location where strong evaporation takes place. However, the main difference is that, for the large injector orifices as considered in this study, the expected lift-off height during the steady injection period is shorter with respect to the liquid length [24]. Therefore there is a region with burning droplet, which is not the case with smaller injector orifice diameters of about 100 µm. As a consequence, a larger amount of soot formation is expected. This effect is Figure 10 compares the computed and experimental shadow images at 5 ms after SOI for the 90 bar case. Note that the experimental image is not available for the entire domain. Images of the different windows at this moment are superimposed [10]. In both cases the spray is dense. Simulation over-predicts the length of the spray. The influence of the fuel quality has a strong impact on the spray evolution and evaporation. For the same conditions of the former figure, distribution of the Sauter Mean Diameter (SMD) for HFO and light Diesel fuel are shown (Figure 11). In the first 100 mm from the nozzle, droplet size distribution is similar for both fuels, although SMD for HFO is slightly larger in agreement with observations in [2]. A strong decrease in droplet size is observed at around 50 mm, which is mainly due to secondary droplet break-up. This distance corresponds to roughly 60 d0 and is in agreement with various findings 6 Copyright © 2012 by ASME reported in the literature (see e.g. the review presented in [26]). Since HFO is injected at elevated temperatures to lower the viscosity to values similar in magnitude to that of Diesel, the surface tension and densities of the two fuels are comparable and, as a consequence, similar break-up occurs. As observed in Figure 6, liquid droplets of light Diesel fuel evaporate completely after roughly 170 mm, while liquid HFO continues to penetrate considerably further downstream. the fuel vapor mass fraction deriving from HFO mixture is considerably lower. This is attributed to only a small fraction of lighter components in the liquid fuel on the one hand and, to a larger extent, substantially lowers vapor pressures of the heaver components (several orders of magnitude) on the other. The initial evaporation of the HFO mixture is hence mainly due to lighter hydrocarbon components, which show preferential evaporation properties [5] and are responsible for the ignition of the mixture. Heavier components are vaporized and burned once the flame heats up those components. While considerable uncertainties are likely to be present concerning the composition of HFO in practice, nonetheless numerical simulation can provide qualitative information on the spray and combustion characteristics. Figure 10: Simulated (upper) and measured (lower) shadow images with HFO at 90 bar and 900 K and 5 ms after SOI. Figure 12: Fuel vapor mass fraction of HFO (upper) and light Diesel (lower) with 900 K temperature and 90 bar pressure at 1 ms after SOI. Figure 11: Droplet Sauter Mean Diameter (SMD) distribution for HFO (upper) and light Diesel fuel (lower) at 90 bar and 900 K and 5 ms after SOI. The evaporation characteristics are essential for autoignition and flame propagation behavior. As shown e.g. in [27], ignition occurs at the most reactive mixture fraction. Since the ignition delay however strongly depends also on the time history of the scalar dissipation rate, accurate predictions are only possible if both accurate evolutions of velocity and turbulence fields are available as well as the correct mixture distribution. Figure 12 gives an impression of the fuel vapor availability at 1 ms after SOI, which is representative for the ignition delay at those thermodynamic conditions. As expected, CONCLUSIONS Numerical simulations of fuel injection for a nozzle size characteristic for large two-stroke marine Diesel engines (0.875 mm hole-diameter) have been carried out for a broad range of test cases, including evaporating and non-evaporating conditions, three different ambient densities and two different fuels for a total of nine different operating conditions. Results have been directly validated against experimental data for penetration length and macroscopic spray morphology obtained from high speed shadow imaging. In analogy to the experimental post-processing of the spray shadow images, a novel post-process methodology for distinction between simulated dense and contour spray penetration length has been proposed and applied in this study, by means of which additional information from the spray can be obtained. Grid sensitivity with respect to penetration length under non-evaporating conditions showed the best results with a mesh size of 2 mm, corresponding approximately to twice the injector hole-diameter, which is consistent with recommendations from previous studies for smaller injectors. Under non-evaporating conditions, the penetration length for light Diesel fuel was well captured for all gas densities 7 Copyright © 2012 by ASME considered and good qualitative agreement of the predicted spray morphology with the measurements was found. Under evaporating but non-reactive conditions, the two trends of faster penetration and increasing separation between dense and contour spray when reducing ambient gas pressure were correctly described, indicating a correct representation of the evaporation behavior at different gas pressures; quantitatively, the penetration length were moderately overpredicted. The influence of fuel quality was studied by injecting HFO, which was approximated by a five-component mixture. With the low fuel quality, the penetration lengths were mildly overpredicted, the trend with respect to the ambient gas pressure as well as the spray morphology however well captured. In agreement with the experiment, evaporation was found to be drastically reduced. This is mainly due to the substantially lower vapor pressures for the heavier components which also constitute the largest portion of the fuel. In this study, a first validation of the spray characteristics under conditions relevant for large two-stroke marine Diesel engines was performed, which forms a prerequisite for future efforts towards predicting the flame structure and emission formation. NOMENCLATURE Projected droplet surface area Adrop CFD Computational Fluid Dynamics Injector orifice diameter d0 FIA Fuel ignition analyzer LFO Light fuel oil HFO Heavy fuel oil SCC Spray Combustion Chamber SOI Start of injection TR Transmission ratio VCVCC Visual constant-volume combustion chamber χ Smoothing rate [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] ACKNOWLEDGMENTS The authors thank Beat von Rotz and Kai Herrmann, Wartsila Switzerland Ltd., for providing experimental data, post-processing and fruitful discussion. The present work has been conducted as part of the HERCULES-B project within EC‘s 7th Framework Program, Contract SCP7-GA-2008-217878. Financial support by the Swiss Federal Government (SER Contract number 154269, SFOE grant numbers 103241 and 102688) is gratefully acknowledged. [13] [14] [15] REFERENCES [16] [1] Internation Maritime Organization, "Technical code on Control of Emission of Nitrogen Oxides from Marine Diesel Engines," MP/CONF. 3/35, Annex, October 1997. [17] 8 C. Fink, et al., "Injection Spray Analyses from Mediam Speed Engine using Marine Fuels," ILASSEurope 2008, Como Lake, Italy, September 2008. K. Takasaki, et al., "Combustion characteristics of trouble-making bunker fuel oil," CIMAC, pp 620-629, 2001. D. Struckmeier, et al., "Multi-Component Modeling of Evaporation, Ignition and Combustion Processes of Heavy Residual Fuel Oil," SAE2009-01-2677, 2009. F. A. Goldsworthy, "Computational fluid dynamics modelling of residual fuel oil combustion in the context of marine diesel engines," Int. J. Engine Research, vol. 7, pp. 181-199, 2006. D. Struckmeier, et al., "New Application and Modeling of Low Ignitability Fuel for Marine Engines," CIMAC, paper No. 117, 2010. N. Kyriakides, et al., "Influence of Heavy Fuel Properties on Spray Atomization for Marine Diesel Engine Application," SAE2009-01-1858, 2009. C. Chryssakis, et al., "Combustion Modeling with Heavy Fuel Oil for Large Marine Diesel Engine Application," CIMAC, paper No. 236, 2010. K. Herrmann, et al., "Reference Data Generation of Spray Characteristics in Relation to Large 2-Stroke Marine Diesel Engines Using a Novel Spray Combustion Chamber Concept," ILASS-Europe 2010, Brno, Czech Repuplic, September 2010. B. von Rotz, et al., "Impact of Evaporation, Swirl and Fuel Quality on the Characteristics of Sprays typical of Large 2-Stroke Marine Diesel Engine Combustion Systems," ILASS-Europe 2011, Estoril, Portugal, September 2011. K. Herrmann, et al., "Validation and Initial Application of a Novel Spray Combustion Chamber Representative of Large Two-Stroke Diesel Engine Combustion Systems," ICLASS 2009, Vail, Colorado, USA, July 2009. CD-adapco, "Methodology Star-CD Version 4.16," 2011. K. Herrmann, et al., "Development of a reference experiment for large diesel engine combustion system optimization," CIMAC, paper No. 98, 2007. R. Schulz, et al., "Assessing the Performance of Spray and Combustion Simulation Tools against Reference Data Obtained in a Spray Combustion Chamber Representative of Large Two-Stroke Diesel Engine Combustion Systems," CIMAC, paper No. 247, 2010. K. Y. Huh and A. D. Gosman, "A phenomenological model of diesel spray atomization," Proc. Int. Conf. on Multiphase Flows, Tsukuba, Japan, 1991. R. D. Reitz and R. Diwakar, "Effect of drop breakup on fuel sprays," SAE 860469, 1986. R. Aneja and J. Abraham, "How far does the liquid penetrate in a diesel engine: Computed results vs. Copyright © 2012 by ASME [18] [19] [20] [21] [22] measurements?," Combustion Science and Technology, vol. 138, pp. 233-255, 1998. N. Abani, et al., "Reduction of numerical parameter dependencies in diesel spray models," Journal of Engineering for Gas Turbines and Power-Transactions of the Asme, vol. 130, May 2008. G. Pizza, et al., "Evaporating and non-evaporating diesel spray simulation: comparison between the ETAB and wave breakup model," Int. J. Vehicle Design, vol. 45, pp. 80-99, 2007. S. Kim, et al., "Improved Eulerian-Lagrangian Spray Simulation by Using an Enhanced Momentum Coupling Model," Asme Internal Combustion Engine Division Fall Technical Conference 2009, Proceedings, pp. 433-442, 2010. X. Jiang, et al., "Physical modelling and advanced simulations of gas-liquid two-phase jet flows in atomization and sprays," Progress in Energy and Combustion Science, vol. 36, pp. 131-167, Apr 2010. G. Barroso, et al., "An Extensive Parametric Study on Diesel Spray Simulation and Verification with Experimental Data," SAE 2003-01-3230, 2003. [23] [24] [25] [26] [27] 9 View publication stats G. M. Faeth, et al., "Structure and breakup properties of sprays," International Journal of Multiphase Flow, vol. 21, pp. 99-127, Dec 1995. D. L. Siebers and B. Higgins, "Flame Lift-Off on Direct-Injection Diesel Sprays Under Quiscent Conditions," SAE2001-01-0530, 2001. D. R. Tree and K. I. Svensson, "Soot processes in compression ignition engines," Progress in Energy and Combustion Science, vol. 33, pp. 272-309, Jun 2007. B. Schneider, "Experimentelle Untersuchungen zur Spraystruktur in transienten, verdampfenden und nicht verdampfenden Brennstoffstrahlen unter Hochdruck," vol. PhD Thesis, ETH Zürich, Switzerland, Diss. ETH No. 15004, 2003. G. De Paola, et al., "Diesel engine simulations with multi-dimensional conditional moment closure," Combustion Science and Technology, vol. 180, pp. 883-899, 2008. Copyright © 2012 by ASME