Spray G - Program 9:30 Introduction Topic 8 – Spray G Nozzle Geometry and Internal Nozzle Flow Organizer: Chris Powell (Argonne) 9:40 Nozzle Geometry and Near-Nozzle Experiments Daniel Duke (Argonne) 10:20 Internal and Near-Nozzle Flow Modeling Ron Grover (GM) 11:00 Break Topic 9 – Evaporative Spray G Organizer: Daniel Vaquerizo (CMT) 11:30 Macroscopic Spray Visualization and Modeling Josh Lacey (Melbourne) 12:10 Spray Mixing Experiments and Modeling Lama Itani (IFPEN) & Lyle Pickett (Sandia) 12:50 Discussion Period 13:00 Lunch September 5th, 2015

ECN 4 Topic 9: Evaporative Spray G Organizers: Joshua Lacey University of Melbourne Daniel Vaquerizo Sánchez CMT – Motores Térmicos

Presentation Contents Experimental Setup and Techniques Simulation Techniques ECN 4 Experimental Results Experimental Summary ECN 4 Simulation Results Simulation Summary September 5th, 2015

Experimental Contributions Three institutions contributed experimental results Istituto Motori CMT – Motores Térmicos University of Melbourne September 5th, 2015

Simulation Contributions Two institutions contributed simulation results Politecnico di Milano Argonne National Laboratory September 5th, 2015

Previous Contributions Other institutions that contributed to ECN 3, whose data is used here for comparisons General Motors Institut Français du Pétrole Energies nouvelles Sandia National Laboratories University of Wisconsin - Madison September 5th, 2015

Spray G Experiments

ECN 3 Experimental Techniques Liquid phase DBI: SNL, IFPEN, UM Mie-scattering: SNL, GM, IM Vapor phase Schlieren: SNL, GM, IM General Motors Sandia National Laboratory September 5th, 2015

ECN 4 Experimental Techniques University of Melbourne Liquid phase DBI: UM Mie-scattering: UM, IM Vapor phase Schlieren: UM, IM, CMT Liquid and vapor phases are measured simultaneously; frame straddling technique used by UM and IM Istituto Motori CMT – Motores Térmicos C-Mos Camera Knife-edge 15° Off-axis mirror f = 516 mm f = 516mm Vessel Lens f = 90 mm Lens f = 200mm f = 50mm Schlieren LED Mie LED Pin-hole Heated plate September 5th, 2015

ECN 4 Experimental Techniques Spray G1 condition of IM and UM setups IM uses a heated plate to achieve high in-chamber temperatures UM has cartridge heaters mounted in the metal of the spray chamber to uniformly heat the walls Istituto Motori University of Melbourne September 5th, 2015

Experimental Results Spray G1 Liquid measurements from Mie-scattering IM liquid penetration extends further into chamber as there is less evaporation Early part of injection event is similar across all experimental setups ECN 3 Liquid Lengths ECN 3 Vapor Lengths September 5th, 2015

Experimental Results Spray G3 Liquid lengths were measured with Mie-scattering As expected, the chief difference between the 20˚C and 60˚C condition, is a decreased liquid residence time, as more fuel is vaporized Early behavior, while the injector is open, is similar Most of the deviation occurs well after the injector is shut Little change in maximum penetration length Vapor profiles lie on top of one another September 5th, 2015

Experimental Results Spray G2 As before, the increased chamber temperature does not affect maximum liquid penetration length, but does impact residence time Maximum liquid penetration is similar to 1 bar, non-flashing case, with residence time reduced Vapor penetrates further and at a different rate than G3 (more vaporization with flashing spray) September 5th, 2015

Mie-scattering and DBI Both are used to measure liquid phase, but each employs a fundamentally different approach Mie-scattering exploits the refraction of light through liquid droplets; the signal is the refracted light (lighter pixels) DBI is a light extinction method in which the dense medium blocks the background illumination; the image formed of the spray is the blocked light (darker pixels) Mie-scattered image DBI image September 5th, 2015

Mie vs. DBI Post-Processing Because there is no “reference” intensity for Mie (black background), we must assign some light level threshold to determine the end of the liquid DBI is “self-calibrating” as the light background provides a reference intensity; the end of the liquid is denoted by a sharp cut off September 5th, 2015

DBI Dependence on Experimental Apparatus In practice, the post-processing of DBI still requires some finesse Optical depth profiles of a spray subjected to the same conditions can vary from one setup to another (because of illumination, collection angle, etc.) Evaporative conditions are susceptible to uncertainty, as the presence of vapor will steer some light (which appears as darker pixels on the image), which impacts the determination of the liquid location As a rule of thumb, use a threshold of I/I0 = 0.9 to denote the liquid boundary Essentially, any light that misses the camera for any reason (blockage by dense medium, small collection angle, vapor scattering the light) results in a signal that appears dark on the image and is treated as liquid spray September 5th, 2015

Spray G DBI versus Mie-scattered Liquid Lengths Although DBI profiles differ from Mie at Spray G1 and G3 (the most and least evaporative conditions), these differences are not dramatic, even for G1 DBI is the preferred method of ECN, but Mie-scattering is probably still useful in establishing the bounds of liquid length at a given condition (and serves as a good “sanity check” when considering the DBI results from a given setup) September 5th, 2015

Summary of all Liquid Measurements for Spray G1; 6 bar, 300˚C The early portion of the injection event is consistent across all labs and techniques (up to ~700-800 µs); liquid edges are sharp here, so there is less experimental uncertainty and better Mie/DBI agreement Significantly more uncertainty after the injector closes; setup and post-processing differences are more apparent in this region Overall, most of the experimental datasets are consistent for all of the setups, and the two different techniques September 5th, 2015

Spray G Liquid Angle Nominal cone angle is apparent early in the injection event, although it falls off in some cases for Spray G1 (SNL, UM) G2/G3 conditions are given at 20˚C chamber temperature Liquid angle increases at conditions with less vaporization (θG1 < θG2 < θG3) For a single-hole injector we might expect the G2 condition to have the highest spray angle; multi-hole plume-to-plume flashing-boiling interaction is more complex September 5th, 2015

Comparison of Liquid and Vapor Angles Vapor and liquid angles at conditions G2/G3 differ slightly at the beginning of the injection event; there is higher variation for G2, where more vaporization occurs Spray G1 has a significant difference between liquid and vapor envelopes due to substantial evaporation September 5th, 2015

High Speed Movies of Spray G2 and G3 Movies showing G2 and G3 conditions on the left and right, respectively Mie-scattered liquid is traced in red, Schlieren measured vapor is traced in blue There are relatively similar spray evolutions at these conditions, though some plume interactions are visible at 0.53 bar 0.53 bar 1 bar September 5th, 2015

Flash-boiling Spray G2 Plume-to-plume interaction is apparent at the 0.53 bar condition Idle condition in an engine might have a lower pressure and lower superheat degree of fuel further 0.53 bar 1 bar September 5th, 2015

Experiments Summary Established a wealth of results for the G1 condition Flash-boiling, while present at the G2 condition, is not severe The maximum liquid penetrations observed are not significantly different when the chamber pressure is dropped from 1 to 0.53 bar (residence time changes and liquid and vapor penetration rates differ) We anticipate that increasing the spray superheat degree or decreasing the cone angle will influence the plume-to-plume interaction September 5th, 2015

Spray G1 Simulations

ECN 4 Simulation Techniques ANL CONVERGE LES: dynamic structure model for turbulence “Blob” injections, KH-RT breakup 1 mm baseline cell size with a 0.125 mm minimum Polimi OpenFOAM with LibICE RANS flow solver, k-ε model for turbulence, Lagrangian approach Huh injection, Pilch-Erdman secondary breakup 4 mm baseline cell size with a 1 mm minimum September 5th, 2015

Review of CFD Penetration Measurement Specifications Liquid Penetration 0.1 % Liquid volume fraction threshold Vapor Penetration 0.1 % Mixture fraction threshold Measured along injector axis Zero point is at the injector tip Penetration Length Side View September 5th, 2015

Vapor Simulation Results Vapor length simulations agrees closely with experiments up to 1000 µs ASI Overall evolution of vapor (curve shape) is similar between both experimental measurements and simulations September 5th, 2015

Liquid Simulation Results Maximum liquid lengths match closely to to those from DBI data The spray behavior of the simulations underpredicts penetration length during a portion of the time the injector is open Non-aligned spray grid increases cone angle and reduces penetration Evaporation models are also a likely contributor Saha and Som, USCAR, 2015 September 5th, 2015

PDI Experimental against Modeling – Radial Scan 15 mm GM Experiments Polimi Simulations September 5th, 2015

PDI Experimental against Modeling – Transverse Scan 15 mm GM Experiments Polimi Simulations September 5th, 2015

Droplet Size Distributions Radial scans are shown to the right at two points in time ASI Experimental droplet sizes are relatively constant Simulations have a wide range of droplet diameters across the radial scan, particularly late in the injection event September 5th, 2015

Experimental Penetration Rates at Spray G1 Conditions Liquid penetration rates are generated from the time derivatives of penetration length Simulation axial velocities are slower overall when compared to the experimental results; some of the high-level behaviors with location and time are captured Radial Scans Transverse Scans September 5th, 2015

Simulation Summary Modeling tools are well-developed and clearly capture the complex behavior present in GDI fuel sprays There is the potential to perform parameter traversals to help better assess the experimental observations (dual-plume cuts for flashing cases, PDI simulations to assess the changes in axial versus radial velocities as flashing-boiling increases) September 5th, 2015

Backup slides September 5th, 2015

ECN 4 Spray G Conditions Three conditions which correspond to different parts of the operating map of a GDI engine Spray G1 Stratified mixture, late injection Spray G2 Throttled, early injection Flash-boiling of iso-octane at this condition Spray G3 Unthrottled, homogeneous mixture, early injection Fuel temperature is maintained at 90˚C in all cases September 5th, 2015

Injection Temperature ECN 4 Spray G Conditions Experimental GDI injector and driver provided by Delphi 8 symmetrically spaced nozzle holes 165 µm hole diameter 780 µs total injection time, at all conditions Nominal spray cone angle of 80° Fuel is iso-octane Two ambient temperatures are listed for Spray G2/G3, because much of the data was taken for the “old” conditions of interest; some comparisons between the two conditions will be presented, but most of the data shown is for the 20˚C condition Spray G1 Spray G2 Spray G3 Injection Pressure 200 bar Injection Temperature 90°C (363.15 K) Ambient Temperature 300°C (573.15 K) 20/60°C (293.15 K) Ambient Density 3.5 kg/m3 0.6 kg/m3 1.1 kg/m3 Nitrogen Pressure 6 bar 0.53/0.5 bar 1 bar Injected Quantity 10 mg September 5th, 2015

Review of PDI and Spray Plume Scans The next few slides provide a quick overview of the relevant geometry for PDI scans and the study of individual spray plumes September 5th, 2015

Along a radial line through the spray axis center point Radial Cross-section Along a radial line through the spray axis center point Zero at injector axis, positive outward Radial Line 15 mm Side View Top View September 5th, 2015

Transverse Cross-section Across the spray perpendicular to the radial line Zero at spray center point Experiments put this at 10 mm radial distance Positive direction clockwise as viewed from injector 10 mm Spray Plume Center 15 mm Transverse Line Side View Top View September 5th, 2015

Dual-Plume Cross-section Taken along a line connecting the center-points of two neighboring spray plumes Zero at center of counter-clockwise-most (as viewed from injector) plume Positive towards second plume (clockwise as viewed from injector) 15 mm Dual-plume Line Side View Top View September 5th, 2015

High Speed Movies of Spray G2 and G3 Movies showing G2 and G3 conditions on the left and right, respectively Mie-scattered liquid is traced in red, Schlieren measured vapor is traced in blue There are relatively similar spray evolutions at these conditions, though some plume interactions are visible at 0.53 bar 0.53 bar 1 bar September 5th, 2015