Preliminary Testing of Sprays with Continuously Variable Spray Geometry

Transcription

1 ILASS-Americas 23rd Annual Conference on Liquid Atomization and Spray Systems, Ventura, CA, May 2011 Preliminary Testing of Sprays with Continuously Variable Spray Geometry J.P. Lien, J. Zhang, T. Fang, G. D. Buckner Department of Mechanical and Aerospace Engineering North Carolina State University, Raleigh, NC Abstract Several recent studies have examined variable geometry spray (VGS) fuel injectors as a means to enable advanced combustion modes (HCCI, LTC, etc.). Many HCCI implementations combine advanced injection timing with conventional late-cycle injections. Optimal mixing, however, requires that spray geometry vary along with injection timing, which cannot be accomplished with traditional injectors. Partial solutions to this problem have been proposed in the form of dual-mode injectors, or injectors which use properties of the fuel flow (such as pressure or cavitations) to influence the spray geometry. The authors propose a design that offers continuous control of the spray angle which is independent of pressure and flow rate. This paper presents the preliminary results of spray and droplet size testing on two prototypes. It is shown that continuous control of the spray angle can be achieved over the desired range, and that spray angle is independent of fuel pressure. Corresponding Author: tfang2@ncsu.edu

2 1 Introduction Designers of combustion entines are faced with an ongoing, critical need to reduce emissions in order to meet increasingly strict regulatory requirements. The US Environmental Protection Agency [1] (among others) have set maximum acceptable levels of particulate matter (PM, or soot) and nitrogen oxides (NO x ) for diesel engines which are difficult to achieve with conventional diesel combustion methods. Absolute improvements in these emissions products havebeen limited by a fundamental PM- NO x trade-off (Fig. 1)[2]. Because reactants in conventional diesel combustion are only partially premixed, many reactions take place in fuel-rich regions. Combustion in such regions gives rise to soot particles due to the scarcity of oxygen. To reduce PM, combustion temperature can be increased (promoting soot oxidization), but higher temperatures increase dissociation of diatomic nitrogen (N 2 2N), resulting in increased NO x formation. Figure 1: Plot of equivalence ratio vs. temperature of combustion, illustrating regions of soot and NO x formation, and the trade-off faced by conventional diesel cumbustion (hollow arrows).[3] Novel combustion modes such as Homogeneous Charge Compression Ignition (HCCI) and Premixed Charge Compression Ignition (PCCI) have been proposed to escape this trade-off by combining a premixed charge (as in spark-ignition (SI) engines) with compression-ignition (CI), as in diesel engines[4]. This is ilustrated by the filled-arrows in Fig. 1, as opposed to the hollow arrows representing conventional disel combustion. The question of how to achieve HCCI combustion in a practical diesel engine, however, remains an open research area. The primary obstacle to implementing HCCI is the wide range of operation (i.e. idle to full-load) required by on-road engines[5]. Because HCCI uses a lean mixture to minimize sooting, it is difficult to achieve power densities comparable to conventional diesel combustion[6]. Dual-mode strategies have been proposed to overcome these challenges by using HCCI under partial load and conventional diesel combustion when full power is required[7, 8, 9]. Because most engine operation takes place at partial loads, the overall emissions are reduced, but greater power is available when needed (e.g. for acceleration). Multi-mode combustion concepts are difficult to implement with conventional fuel injectors, however, as their fixed spray geometries cannot be optimal for all modes at once [10]. To address this problem, our research group is currently developing a novel fuel injector that can provide continuous spray angle control without modulating fuel flow rate or any other vital injection parameters. The injector design is described in Section 2. Section 3 discusses the experimental techniques used to assess the spray characteristics, while Section 4 presents the results of preliminary spray tests conducted with two early prototypes. 2 Injector Concept The injector and its associated control system are described in detail in [11]. This section discusses aspects of the design related to the spray nozzle. The goals of the design were based on the work of Fang in [10]. Assuming a centrally-positioned injector with a spray angle that targets the lip of the piston bowl as the piston advances throughout the compression stroke, the optimal spray angle as a function of time can be computed from the piston and cylinder geometry. The apray angle trajectory for a representative piston and cylinder is shown in Fig. 2. Another critical design parameter for fuel injectors is the size distribution of droplets in the fuel spray[12, 13], which depends on the size and geometry of the nozzle outlet, as well as the pressure and chemical properties (e.g. volatility) of the fuel. This gives rise to two secondary design objectives for the VGS nozzle: to minimize the variation in droplet size over the full range of angles, and to keep the spray s Sauter Mean Diameter (SMD) less than 50 microns. An externally-opening pintle-type nozzle was selected as the mechanism to achieve the desired variation in spray geometry. By adjusting the displacement of the pintle from the nozzle orifice, the included angle of the hollow cone spray can be varied over a large range. The design was refined though CFD simulation, rapid prototype testing, and by 2

3 (a) (b) Figure 2: (a) Spray angle targeting piston bowl lip. (b) Desired injection angle as a function of crank angle. comparison to existing variable-spray nozzles (Fig. 3). Through this process a step-type pintle was determined to maximize the achievable cone angle, and the shape of the cavity behind the spray nozzle was improved. Because the size of the annular gap between the pintle and nozzle tip is closely related to the size of the droplets in the resulting spray, this parameter was used to characterize the scale. Based on available data for commercial pintletype nozzles it was determined that a 100µm annular gap represented a reasonable scale for a fuel injector prototype. In order to limit manufacturing time and costs, however, a 2x scale prototype was fabricated for preliminary testing. A nozzle profile with a 200µm annular gap was simulated using AN- SYS CFX to investigate the relationship between pintle position ans spray angle. The CFD simulations used an Eulerian multiphase model with two phases. Room temperature air and liquid water were chosen as the reference fluids since those would be used for the initial experiments. Since CFX only simulates 3D geometries, a single layer of mesh elements was extruded from a 2D cross-section with thickness selected so as to produce reasonable aspect ratios. User controls were applied during the meshing process to specify a fine mesh in the narrowest areas of the injector. All simulations were steadystate, which required the use of velocity boundary conditions on the inlet. Pressure boundary conditions were a more natural choice but did not produce convergence using the steady-state solver. Simulations were performed at a variety of inlet velocities, with low velocities more likely to converge without bootstrapping. For each pintle position a lowvelocity (typically 0.1 m/s) simulation was initialized with an empty control volume, and run until it converged (as measured by RMS residual magnitudes) to a steady-state. This solution was used as an initial condition for simulations with higher inlet velocities. Once each solution reached steady state, the pressure at the inlet could be computed. This process was repeated until an inlet pressure close to the target (1000psi) was obtained. Figure 3: 2D axisymmmetric CFD simulation of an early nozzle design in intermediate position. Further simulations reproduced the full range of angles by movement of the pintle. In order to validate the CFD model, the spray angle and droplet size of the 2x prototype were measured experimentally using pressurized nitrogen and water as a working fluid. These experiments and their results are described in detail in the following sections. On the basis of these results, it was determined that a smaller orifice and pintle were needed to reduce the droplet size to more realistic levels and facilitate further testing at higher pressures. A 100µm annular gap with a 1.5mm pintle diameter was chosen to be consistent with commercial scale injectors (a 1x prototype). The reduced orifice area was expected to produce a spray with smaller droplets. A second goal was to build a pro- 3

4 totype with standard exterior dimensions that could interface with a constant volume combustion chamber for further experimentation. Thus a P-type diesel direct injector was used to determine the external dimensions of the nozzle. Overall dimensions of the injector were significantly reduced to lower material costs and produce a more realistically sized device. The spray and droplet size tests were repeated, along with further CFD simulations. In addition, this prototype was fitted with a piezoelectric linear actuator to demonstrate electronic control of the pintle position and the ability to dynamically vary the spray during the injection. 3 Experimental Methods The primary goals of the spray experiments were to confirm to confirm the spray angle/pintle position relationship predicted by the CFD simulations, and to measure the size of the spray droplets. A small tank of working fluid (either water or iso-octane) was statically pressurized (up to 100bar) using a nitrogen bottle. Flow was controlled using an external solenoid valve. A Phantom high-speed camera (Vision Research, Wayne, NJ) was used to capture images of the spray. From these images the included angle of the spray cone was measured by manually approximating it with tangent lines in a drawing program and then computing the angle between those lines. This process is illustrated in Fig. 4. Some spray images are shown with corresponding CFD simulations in Figure 5. (a) Figure 4: (a) Frame from spray video, (b) Frame with angle measured. The size of droplets in the spray was measured using a Malvern Instruments (Malvern, Worcestershire, UK) Spraytec laser diffraction instrument. The Spraytec emits a beam of coherent light which passes through and is scattered by a suspension of transparent droplets in air. The beam is then measured by an array of pickups on the far side. A Miescattering model relates the intensities observed by (b) the pickups to the size of the droplets in the sample. Because the hollow-cone spray is likely to scatter the beam multiple times, the Spraytec also incorporates a proprietary multiple-scattering model. The spray droplets typically have irregular shapes, so some care must be taken in expressing their size. The Sauter Mean Diameter (SMD) was used as a summary statistic for each spray. 4 Preliminary Results Observed and simulated spray angles for the 2x prototype are plotted against pintle position in Figure 7a. The data appear to show that the CFD simulations predicted the relationship with reasonable accuracy, and that the pressure of the injected fluid has a negligible impact on the spray angle within the tested range. Spray tests on the 1x prototype were performed as before, but using isooctane (2,2,4- Trimethylpentane) as the working fluid in place of water. The 1x prototype demonstrated increase in the range of spray angles and a decrease in the length of pintle travel required to achieve them. The increased range brings the injector closer to the desired 160 degree maximum angle, and the reduced stroke simplifies design of the positioning actuator. Plots of spray angle data for the two prototypes are shown in Fig. 7. The laser-diffraction measurements on the 2x prototype showed a downward trend in droplet size with respect to pressure, which was promising since the experiments were run at up to 100bar, only one tenth of typical direct injection pressures. The relationship of spray angle to pintle position is more complicated, and the data are too sparse to make any firm inferences about it. Figure 6 shows both of these trends as computed from these experiments. The droplet experiments on the 1x prototype showed a significant reduction in Sauter Mean Diameter over teh 2x device, but the iso-octane working fluid prevented the same range of measurements used previously. For injection pressures above 300 psi the spray absorbed or reflected over 90% of the light emitted by the diffraction instrument, making accurate measurements of droplet size impossible. Measurements were obtained at some pintle positions with a 500 psi inlet pressure, but at higher pressures too little data was available to construct a meaningful curve. From the data available, shown in Fig. 8, it can be seen that the the droplet size is below 50 µm with 500 psi inlet pressure at most pintle positions (but not the widest or narrowest). One can also infer that as pressure increases, droplet size will be reduced for most, if not all, spray angles. 4

5 Figure 5: Still frames from spray videos (left) with corresponding CFD simulation results (right). CFD plots are colored according to flow velocity, lighter colors indicate higher velocity magnitude. High speed video was also used to capture electromechanical actuation of the final prototype using a PZT stack actuator. For these tests an open loop input consisting of an 800V 70Hz square wave was used. Still images from this video are shown in Fig. 9. Further details about the design, as well as closedloop electromechanical control are discussed in [11]. 5 Conclusions This paper has described the design and testing of a nozzle for a variable geometry spray fuel injector. The final (1x scale) prototype produces hollow-cone sprays with included angles from degrees, very close to our design goal of degrees. By aggressively reducing the orifice size, we believe we have achieved our design goal of a fuel spray with droplets smaller than 50µm SMD, although experimental difficulties make this somewhat uncertain. Although the design will require future improvements (most notably to incorporate an integral flow control mechanism actuator), the practicality of the VGS concept has been demonstrated. Acknowledgements The authors wish to thank Brian Owen and Skip Richardson for fabricating the injector prototypes, accessories, and test fixtures. This work was supported by the National Science Foundation through NSF grant CBET References [1] Federal Register, 66: , [2] T. Kaminoto and M. Bae. SAE transactions, 97: , [3] S. Kook, C. Bae, P.C. Miles, D. Choi, and L.M. Pickett

6 [4] Mingfa Yao, Zhaolei Zheng, and Haifeng Liu. Progress in Energy and Combustion Science, 35: , [5] J.E. Dec. Proceedings of the Combustion Institute, 32(2): , [6] Magnus Christensen. PhD thesis, Lund University, [7] Bertrand Gatellier, Bruno Walter, and Marjorie Miche. In Pierre Duret, editor, A new generation of engine combustion processes for the future?, pp Editions Technip, Paris, [8] Kevin Duffy. Diesel Engine Emission Reduction Conference, [9] Yong Sun and Rolf D. Reitz. SAE 2006 World Congress & Exhibition, April SAE Paper [10] Tiegang Fang, Robert E. Coverdill, Chia fon F. Lee, and Robert A. White. Fuel, 87: , [11] J P Lien. PhD thesis, North Carolina State University, [12] J. Yang, E.W. Kaiser, W.O. Siegl, and R.W. Anderson. SAE Transactions, [13] M A Patterson, S-C Kong, G J Hampson, and R D Reitz. SAE transactions, 103(3): ,

7 mm 0.2mm 0.3mm 0.4mm 0.5mm psi 1000psi 1200psi 1400psi 1500psi Sauter Mean Diameter (mm) Sauter Mean Diameter (mm) Inlet Pressure (psi) (a) Relative Pintle Position (mm) (b) Figure 6: (a) Droplet SMD vs. fluid pressure, (b) Droplet SMD vs. pintle position. Both results are for the 2x scale prototype psi 500 psi 800 psi 1500 psi CFD (1000 psi) psi 1000psi 1500psi Spray Angle (deg) Spray Angle (deg) Pintle Position (mm) Pintle Position (mm) (a) 2x Prototype (b) 1x Prototype Figure 7: Spray angle vs. pintle position, data from CFD simulations and experiments. 7

8 90 300psi 500psi 80 Sauter Mean Diameter (µ m) Relative Pintle Position (mm) Figure 8: (a) Droplet Sauter Mean Diameter vs. pintle position at two inlet pressures, (b) Histogram of droplet diameters at 500psi, 0.45mm pintle position. (a) 0 ms (b) 8.4 ms Figure 9: Frames from video of electromechanically actuated PZT, using an 800V, 70Hz, square wave input. 8