Experimental Investigation of Atomization Behavior of Renewable Biofuels

Transcription

1 ILASS Americas 26th Annual Conference on Liquid Atomization and Spray Systems, Portland, OR, May 2014 Experimental Investigation of Atomization Behavior of Renewable Biofuels A.G. Silver and V.G. McDonell 1 UCI Combustion Laboratory University of California, Irvine Irvine, Ca USA Abstract This paper presents results from an experimental investigation of the macroscopic and microscopic atomization behavior of B99 biodiesel, ethanol, B99-ethanol blends, methanol, and an F-76-Algae biodiesel blend. In addition, conventional F-76 and Diesel #2 sprays are characterized as a base case to compare with. The experimental spray apparatus consists of a plain air-blast atomizer mounted on a traverse. The breakup characterization of each fuel is conducted under the same air flow conditions in order to simulate use as a drop in source to existing gas turbines with only slight modifications. For this study, a Phase Doppler Particle Analyzer system is employed to gain information on drop size, SMD, velocity, and volumetric flux distribution across the spray plume. A Vision Research Phantom high speed camera is used to gather high speed cinematography of the sprays for use in observing breakup characteristics and providing additional insight. The results illustrate how the fuel type impacts the atomization and spray characteristics. Future work will entail swirl stabilized combustion of each fuel in order to compare with nonreacting flow results. 1 Corresponding author, mcdonell@ucicl.uci.edu, x 11121

2 Introduction Interest in renewable biofuels is growing due to a number of factors including increased energy demand, the depletion of current fossil fuel deposits, fluctuating petroleum fuel costs, stimulating the agriculture economy, and realization of climate change [1]. Past research has focused mainly on biodiesel and ethanol performance in internal combustion engines [2]. The use of these fuels can be extended to applications in gas turbines, specifically those used in transportation. This extension is partially due to the high energy density associated with liquid fuels. Additional renewable fuels and fuel blends such as methanol and biodiesel-ethanol blends have received attention. However, all of these renewable fuels exhibit different physical properties (viscosity, density, surface, tension, etc.) that affect their atomization performance. Biodiesel displays difficulty during cold start the due to its crystallizing property at low temperatures. In addition, pump and filter life in the fuel system is shortened due to long term storage issues. Even with these weak points, numerous attempts have been made to use biodiesel in engines because of its low emission characteristics [3]. The feasibility of using biofuels in gas turbine systems is heavily dependent on meeting strict air quality standards for carbon monoxide (CO), oxides of nitrogen (NO x ), oxides of sulfur (SO x ), and hydrocarbons (HC) [4],[5]. While studies have been done on biodiesels, emission results are inconsistent. Most studies show increased levels in CO with varying NO x outputs. However, oxygen present in biodiesels is often cited as an important factor in promoting more complete combustion [6]. With improved atomization, gas turbines operating on biodiesels are speculated to have improved emissions as compared to those using conventional diesel. The most limiting factor in biodiesel atomization is the relative increase in dynamic viscosity and surface tension. Both of these fluid properties are heavily tied to atomization behavior in that the increased viscosity and surface tension limit droplet breakup and lead to larger average droplet sizes which in turn increase residence time and NO x formation. Although there are other modes of improving emissions, improved fuel injector design and mixing has been shown to reduce NO x, CO, and HC outputs [7]. Currently, more research is required in order to develop new designs to accommodate and optimize the performance for these relatively new fuels. While more optimal designs are under development, the use of biofuels in existing gas turbines is already taking place. An array of literature regarding the atomization of the biofuels under scope can be found. Park et al. studied the atomization and combustion characteristics of biodiesel-blended fuels in a common-rail diesel engine. They found atomization suffered as more biodiesel was blended with conventional diesel. This lead to required attention to modify the diesel engine by optimizing the injection timing and injection pressure [8]. Suh et al. investigated macroscopic and microscopic characteristics of biodiesel fuel sprays. They suggested that the biodiesel sprays exhibit lower velocity and larger droplet diameters as compared to conventional diesel, and that there is a great difference in the high blending ratio of the biodiesel [9]. In an effort to achieve improved atomization, ethanol is regarded as a viable option for a fuel additive. Ethanol is derived from the fermentation of sucrose from that of either sugar cane or corn. It is a wellestablished and understood biofuel. On its own, ethanol has demonstrated lower emissions, but has only 63% latent heat content of diesel. This requires greater fuel consumption and storage. The decreased emissions from ethanol are partly due to its superior atomization, in that the dynamic viscosity, density, and surface tension are lower than that of diesel. This makes ethanol a good additive to other fuels in helping to improve atomization. Fang et. al shows how ethanol s high volatility and low Cetane number are favorable for low temperature combustion modes, and that an ethanol-dieselbiodiesel blend was effective in reducing NO x and smoke in premixed low temperature combustion [10]. Methanol is another renewable fuel of interest. Its use as a substitute fuel has been demonstrated in gas turbines, and has potential to significantly reduce greenhouse gas emissions. It is produced via four primary pathways: municipal waste, industrial waste, biomass, and carbon dioxide. Methanol too has its tradeoffs. With roughly half the energy density to that of conventional diesels, it is corrosive and thus faces material compatibility issues [11]. Hain tested methanol using a Pratt & Whitney FT4C TWIN-PAC 50 MW e gas turbine. Due to methanol s decreased heating value, the machine was only able to operate at just above half load. However, emissions evaluation displayed 75% reduction in NO x, reduced particulate output, and no SO 2 [12]. These past studies have led to increased interest in renewable fuels, still more detailed information on their atomization performance is still needed to make these options more viable. The current experimental investigation studies the atomization behavior for a variety of biofuels under drop in air flow conditions. That is to say all of the fuels of interest are atomized under the same air flow conditions. The renewable fuels studied are an F- 76/algae derived biodiesel blend, methanol, ethanol, B99, and two B99/ethanol blends. Experimental Setup Under partnership with the UCI Chemical Engineering and Materials Sciences Department and the 1

3 Office of Naval Research (ONR), a burner was developed in order to simulate the effluent from a conventional gas turbine combustor for use of advanced materials testing for turbine blades and coatings. Materials testing required continual operation of the burner for tests lasting from 500 to 1000 hours while using less than a (55-gallon) drum of candidate fuel. The burner operates using a twin-fluid air-blast atomizer (see Figure 1), similar to that used by Rizk and Lefebvre [13]. The atomizer section of the burner is used in this study. Rather than using co-flowing air, a conical section converges the air stream upon a liquid orifice. A centering piece is used to align the flue tube concentrically within the air line. The atomizers geometric parameters are listed in Table 1. Air Fuel Air Figure 1. Twin-fluid air blast atomizer, cross-section view. Parameter Liquid Orifice Diameter Air Orifice Diameter Atomizing Air Cone Angle Table 1. Atomizer parameters mm mm 64.4 degrees The fuels for spray testing were supplied from 5 gallon tanks pressurized with nitrogen. The pressurized fuel was then sent through a filter to remove any particulate, and the volumetric flow rate then metered using a rotometer calibrated for each fuel. This design allowed the fuel tube to be loosened and re-positioned in order to vary the pressure drop across the exit. The upstream air pressure is controlled using a gas regulator and the mass flow rate monitored using a rotometer which was calibrated using a laminar flow element. A mercury manometer recorded pressure drop readings upstream of the air-blast atomizer. These measurements, along with the air mass flow rate, were then used to calculate the atomizing air velocity and the effective area of the 2 contraction section of the atomizer, via the Bernoulli equation. For each fuel tested, the upstream air pressure is maintained at 80 psi. The fuel was then subjected to three atomizing air flow velocities according to the pressure drop. These target points are listed in Table 2. By varying the atomizing air downstream effective area, the air mass flow rate changed accordingly. Parameter Low Mid High DP (kpa) Average Air Flow (g/min) Table 2. Atomizing Air Parameters. Fuels With the main variant in this study being fuel type and blend, a total of 8 fuels are investigated. To establish baseline performance, F-76 naval distillate fuel (F76) and conventional #2 diesel fuel (DF2) are used. These fuels were provided by the Office of Naval Research and UCI North Campus respectively. ONR also provided an F-76/algae-derived biodiesel blend mixed on a 50/50 mass basis (F76-Algae). ONR is interested in implementing renewable fuel usage in the naval fleet, particularly those derived from algae sources. The F76- Algae blend is similar in physical properties to conventional petroleum fuels, in part due to the significant amount of petroleum fuel in the blend. The production of the algae involved a process called hydrotreatment. This process generally involves the reaction of a triglyceride feedstock (oil from micro-algae) with hydrogen at high pressure and temperature with a catalyst present. Further refinement results in a liquid hydrocarbon fuel that is chemically and physically similar to conventional petroleum fuels [14]. Methanol and ethanol pure fuels are also studied, and were both obtained from the UCI chemical store. Surrogate chemical formulas and thermochemistry data for DF2, F76, F76- Algae, methanol, and ethanol were found from literature [15], [16]. B99, 99% biodiesel and 1% diesel, was obtained from a provider in Temecula, California. Using on campus facilities, the chemical composition of the B99 fuel was determined using Mass Spectrometry Gas Chromatography (MSGC). Two runs were performed in order to ensure quality and repeatability of the results. The runs lasted for 80 minutes and ranged from degrees Celsius in 3 degree increments. The gas chromatograph separates out individual species from the sample by sending the mixture over a packed bed column that has variable affinity for the various components. At the column outlet, a detector quantifies the relative abundance of each species as it elutes off. The mass spectrometer then ionizes the constituents and determines the molecular fingerprint based on the mass

4 Species Chemical Formula Mass Fraction (%) Linoleic Acid, Methyl Ester C 19 H 34 O Oleic Acid, Methyl Ester C 19 H 36 O Palmitic Acid, Methyl Ester C 17 H 34 O Stearic Acid, Methyl Ester C 19 H 38 O Table 3. Fuel types and their physical properties. Fuel Approx. Chemical Formula Density (kg/m 3 ) Kinematic Viscosity (mm 2 /s) Surface Tension (kg/s 2 ) Flow Rate ml/min F-76 Navy Distillate C H DF2 C H F-76/Algae Blend C H Methanol CH 3 O H Ethanol C 2 H 5 OH B99 C H O BE80 C H O BE60 C 27.2 H 59.9 O Table 4. Fuel types and their physical properties. to charge ratios of the ion fragments and spectral libraries stored within the instrument. Thus the MSGC technique identifies the type and concentration of various molecular species within a complex mixture [17]. The resulting composition of the B99 is displayed in Table 4. Lastly, two B99-ethanol blends were created on the following mass bases; one comprised of 80% B99 and 20% ethanol (BE80), and the other 60% B99 and 40% ethanol (BE60). For the eight fuels of interest, the density, surface tension, and viscosity were measured using weighing of a filled graduated cylinder, stalagmometer, and ball drop viscometer, respectively. The measured properties and approximate chemical formula are displayed in Table 3. With the experimental spray rig designed to flow roughly 4 ml/min of F76, this equates to a theoretical power output of 2.4 kw. The volumetric flow rate for each fuel is adjusted in order to provide the same power output in order to simulate matching demand levels for a real turbine. The calculated volumetric flow rates are additionally displayed in Table 4. Military specifications for F76 distillate require the kinematic viscosity to reside between 1.7 and 4.3 mm 2 /s at 40 o C [18]. Using graphs depicting kinematic viscosity limits versus fuel temperature for aviation fuels [19], this kinematic viscosity limits for F76 are approximated for a measurement temperature of 20 o C. These new limits require the kinematic viscosity to range roughly between 2.4 and 5 mm 2 /s. It is seen that the F76, DF2, F76-Algae, and the BE60 blend all meet the kinematic viscosity requirements. The BE80 is just outside of the maximum allowable range, and the B99 is much too viscous to meet the requirement. On the lower limit, ethanol and methanol s low viscosities do not enable them to meet the standards at ambient conditions. The maximum density specified for F76 distillate at 15 o C is 876 kg/m 3, which then after correcting by the same procedure to 20 o C, becomes approximately kg/m 3 [18],[19]. It is seen that all fuels but the B99 meet this density standard. There is no existing F76 distillate standard for surface tension. Measurement Methods Two optical methods were used to characterize the atomization performance for each fuel. The first employed the use of a Phase Doppler Particle Analyzer (TSI FSA-4000) to capture droplet size data, coupled with a Laser Doppler Velocimeter (LDV) system to record droplet velocity. The PDPA/LDV system consists of focusing an argon-ion laser into a fiber optic transmitter. The beam is then split into three wavelengths: green (514.5 nm), blue (488 nm), and violet (476.5 nm). This experiment used only the green and blue beams to capture 2-D data. The transmitted laser beams, 2 green and 2 blue, intersect to create a sample volume from which measurements are taken. As droplets pass through the sample volume, the resulting laser infringement pattern is refracted through the droplet and scattered. A receiver unit positioned at 30 o off axis of the transmitter collects the refracted light and converts it into electrical signals via photomultiplier tubes (PMTs). The frequency of this signal, known as the Doppler frequency, is measured and used to calculate the droplet velocity. The particle size is determined by the measured spatial frequency, or spacing between the scattered fringes at the collecting optics. The PDPA/ 3

5 LDV system is able to calculate a variety of useful parameters that assist in characterizing the spray plume such as Sauter Mean Diameter (SMD), average diameter (D10), and 2-component volume flux and velocity [20]. The second optical measurement method included a Vision Research Phantom 7.1 high-speed camera used to capture high speed shadowgraphy. These images were used to help visualize the breakup process of each spray. A schematic of the experimental setup is presented in Error! Reference source not found.. Figure 2. Experimental Setup Layout. higher resolution in the densest region of the spray. Data runs lasted until a minimum of 50,000 samples were recorded in order to provide a sufficient average of the data. This requirement excluded some data points further than 5 mm from the centerline. Figure 3. displays a still from the high speed imaging with the appropriate scaling and measurement locations. Results and Discussion During the initial testing and experimental design stages, high speed video and PDPA/LDV data were observed. High speed images displayed the breakup process taking place. The air blast nozzle geometry allows for the co-flowing air to atomize the majority of the fluid before it exits the nozzle. There still appears to be some buildup of fluid at the nozzle exit, which is then shed off and the ligaments break up into individual droplets. This phenomenon occurred more with higher viscosity fuels and lower air to liquid ratios (ALR). Stable atomization occurred further downstream; as such, PDPA/LDV data were chosen to be taken at an axial distance 10 mm downstream of the nozzle exit. At this location, the fuel sprays spanned approximately 5 7 mm radially from the nozzle centerline. PDPA/LDV point measurements were taken 1 mm apart in the outer region of the spray, and.5 mm apart between ±2 mm of the centerline in order to provide 4 +Y +X Figure 3. Still image of F76 spray plume with appropriate scaling and PDPA/LDV traverse location.

6 Velocity (m/s) SMD (um) With the air control parameter being the pressure drop across the nozzle, the atomizing air velocity can be controlled. Figure 4. below depicts the three pressure drop trials effects on velocity for F Radial Distance From Nozzle Tip (mm) Figure 4. Velocity profiles for F76 at three pressure drops: low (red), middle (green), and high (blue). Two components of velocity are displayed with depicting line styles: the axial velocity (solid), and the radial velocity (dashed). The LDV system is able to capture the velocity profile of the flow. The average droplet velocity appears normally distributed about the centerline. The maximum axial velocity should exist exactly along the centerline, however it is slightly skewed. Similarly for the radial velocity, one presumes the droplets to have no horizontal movement, but it is seen that this occurs 2 mm to the right of the nozzle. Both these offsets could be attributed to the spray deflected to one side, or error in alignment of the traverse mechanism. As predicted by Bernoulli s equation, the droplets axial velocities increase with the pressure drop. This normal velocity distribution and scaling magnitude with pressure drop is observed for all fuels. The SMD behavior for each test case is also obtained across the spray plume. Figure 5. displays this distribution for the three F76 test cases. It is readily seen that as increased atomizing air velocity results in smaller droplet sizes. This result it predicted by Lefebvre s SMD equation [21] for an air blast atomizer ( ) ( ) ( ) ( ) ( ) 5 where d o is the characteristic length. The value used for d o is obtained from the effective area for the atomization air inside the nozzle. In addition the ε term introduced is a modifying constant to account for differences in atomizer geometry. For this particular atomizer, ε = Radial Distance From Nozzle Tip (mm) Figure 5. SMD profiles for F76 at three air pressure drops: low (red), medium (green), and high (blue). The SMD distribution curves scale inversely with atomizing air velocity, leading to improved atomization. For each pressure drop, SMD increases in the region close to the centerline. With the air circuit converging on the fuel stream, the outer layers of the liquid jet are the first subject to shearing forces. This effect suggests that the majority of the liquid located within ±2 mm of the centerline does not experience as much momentum transfer from the air to liquid. The SMD values across the spray might possibly be made more uniform by increasing the ALR, thus introducing a larger momentum flux into the atomization process. With the geometry scaling factor taken into account, theoretical SMD values predicted by equation (1), as Fuel ALR Theoretical SMD ( m) DF F F76-Algae Methanol Ethanol B BE BE Table 5. SMD and experimental ALR values for all fuels at high pressure drop.

7 SMD (um) DF2 F76 F76-Algae Ethanol Methanol B99 BE80 BE Figure 6. SMD Radial distributions Position for From all Nozze eight fuels Tip (mm) for high atomizing air pressure drop. well as the corresponding ALRs are tabulated for the high pressure drop case. Experimental SMD results for all fuels are plotted together on the same graph in Figure 6. The theoretical results from Table 5. can only help predict the SMD distributions to a certain degree. For the low viscosity fuels F76, DF2, and F76-Algae, atomization occurs mainly due to aerodynamic forces [21] and all atomize to produce similar SMD values. The B99 fuel has a clear increase in SMD due to its relatively larger viscosity and surface tension. BE80 and BE60 fuel blends were hypothesized to perform better than the pure B99 due to the decrease in viscosity attributed to the added ethanol. These blends require higher volumetric flow rates and decrease the ALR. Regardless, the two blends performed as predicted by theoretical SMD values, with BE60 atomizing slightly better than BE80. In the outer regions of the spray, BE60 atomizes nearly the same as the conventional fuels, but produces slightly larger droplet sizes in the bulk of the flow. The methanol and ethanol sprays do not behave as predicted by equation (1). These fuels are predicted by theory to perform the best and achieve the lowest SMDs of all the fuel candidates. In order to meet the power output requirements, much larger flow rates must be provided for these fuels. This leads to a significant reduction in ALR, wherein large amounts of the fluid buildup at the nozzle exit and are then shed off in ligaments in a secondary atomization phase. These ligaments are atomized by the air flow following the primary liquid-air momentum exchange occurring near the liquid orifice 6 inside the nozzle, leading to large droplets shed off to the right plane of the spray plume. The ethanol does experience a region of fine atomization to the left; however the larger droplets from accumulation are shed off to the right, increasing in a linear fashion across the spray. Methanol never experiences any region of finer atomization like ethanol and has poor overall droplet break up. These effects for both fuels can be visually observed as shown in Figure 7. These large droplets can cause issues during reacting flow as they meet with the combustion chamber wall and coke on the surface. Excluding those for methanol and ethanol, the images in Figure 7. do not depict the spray plumes to be skewed off axis. Conclusions This study investigated the atomization performance for several renewable and conventional fuels in a twinfluid air blast atomizer. Breakup of each fuel was conducted under the same air flow conditions in order to simulate use as a drop in fuel to existing gas turbines with only slight modifications. High speed images and SMD distributions show that most fuels atomize similarly under the prompt atomization mode inside the air blast atomizer used. For the more viscous fuels, B99, and flows with lower ALR, namely ethanol and methanol, fuels experienced buildup at the nozzle exit and were then shed off in ligaments during secondary atomization. This suggests that methanol and ethanol would not appear to be good candidates as a drop-in fuel source to meet normal power loads. In order to employ

8 Figure 7. High Speed still images for all eight fuels for high atomizing air pressure drop. these pure fuels, the liquid flow rate should be decreased in order to improve atomization. Consequently, this atomization gain comes at the expense of lower engine output levels due to the fuels lower heating values. The BE80 and BE60 fuel blends, with their similar properties to F76, produce slightly larger SMD values than the conventional fuels. This makes them potential candidates to be used as drop-in fuels. F76 and DF2 achieved the best atomization, as expected because the burner was developed specifically for these liquids. The F76-Algae blend performs well, largely due to its large F76 blending contribution. These dropin fuel candidates can help progress the transition to more renewable fuels and lead to future designs tailored for their use. Acknowledgements The authors would like to thank David Shifler from the Office of Naval Research, Bruce Rodman of the Naval Surface Warfare Center, Carderock Division for their support in development of the gas turbine materials testing rig from which this burner was developed. We would also like to thank Professor Daniel Mumm for his design and fabrication of the materials testing rig, Professor Scott Samuelsen for his continual guidance, Adrian Navarez and Justin Legg for their prior contribution to the burner data and performance, Chris Bolszo for his and support and aid in data analysis, and Jamie Ibrahim for helping conduct experiments. Nomenclature F76 F-76 naval distillate fuel DF2 #2 Diesel Fuel F76-Algae F-76/Algae-biodiesel 50/50 blend BE80 B99/Ethanol 80/20 blend BE60 B99/Ethanol 60/40 blend ALR air to liquid mass ratio SMD Sauter mean diameter PDPA Phase Doppler particle analyzer LDV Laser Doppler velocimeter density U velocity surface tension ε SMD modifying constant d diameter Subscripts L liquid A air o characteristic length References 1. Gupta, K. K., A. Rehman, et al. "Bio-fuels for the gas turbine: A review." Renewable and Sustainable Energy Reviews 14(9): A. Demirbas. Progress and recent trends in biofuels. Progress in Energy and Combustion Science, 33(1): 1-18,

9 3. Kim, H. J., H. K. Suh, et al. (2008). "An Experimental and Numerical Investigation of Atomization Characteristics of Biodiesel, Dimethyl Ether, and Biodiesel Ethanol Blended Fuel." Energy & Fuels 22(3): Bolszo, C. D. and V. G. McDonell (2009). "Emissions optimization of a biodiesel fired gas turbine." Proceedings of the Combustion Institute 32(2): Board, C. A. R. (2010). "RULE 1134 Stationary Gas Turbines." Justin M. Legg, A. A. N., and Vincent G. McDonell (2011). Performance of Alternative Liquid Fuels in a Swirl-Stabilized Research Combustor: Bryden., K. W. R. K. M. (2011). Combustion Engineering. NW, Taylor & Francis Group. 8. Lee, C. S., S. W. Park, et al. (2005). "An Experimental Study on the Atomization and Combustion Characteristics of Biodiesel-Blended Fuels." Energy & Fuels 19(5): Suh, H. K.; Park, S. W.; Kwon, S. I.; Lee, C. S. Trans. KSAE 2004, 12, Fang, Q., J. Fang, et al. "Effects of ethanol-dieselbiodiesel blends on combustion and emissions in premixed low temperature combustion." Applied Thermal Engineering 54(2): Karen Law, J. R., Michael Jackson (2013). Methanol as a Renewable Energy Resource: Y. Hain1, B. C., N. Rappoport, Mordechai (Modi) Reshef, S. Baitel. (2012). "Methanol as a low cost alternative fuel for emission reduction in gas turbines." 13. Rizk, N. K. and Lefebvre, A. H., Spray characteristics of plain-jet airblast atomizers. Journal of Engineering for Gas Turbines and Power, 106: (1984) 14. Serrano-Ruiz, J. C., Ramos-Fernández, E. V., and Sepúlveda-Escribano, A., Energy & Environmental Science, 5: (2012). 15. G. Steinfeld, R. Sanderson, H. Ghezel-Ayagh, and S. Abens. Distillate fuel processing for marine fuel cell applications. AICHE spring 2000 meeting, 5 9 March 2000, Atlanta, GA, USA. 16. P.J. Linstrom and W.G. Mallard, Eds. NIST Chemistry WebBook, NIST Standard Reference Database Number 69, National Institute of Standards and Technology, Gaithersburg MD, 20899, Bolszo, C. M., V. Samuelsen, S. (2007). Impact of Biodiesel on Fuel Preparation and Emissions for a Liquid Fired Gas Turbine Engine. ASME Turbo Expo: Power for Land, Sea, and Air, Montreal, Canada, ASME. 18. Department of Defense, Performance Specification Fuel, Naval Distillate, Military Specification MIL-DTL_16884L. 19. International, A. Standard Practice for Qualification and Approval of New Aviation Turbine Fuels and Fuel Additives. D Phase Doppler Particle Analyzer (PDPA)/Laser Doppler Velocimeter (LDV) Operations Manual. TSI Incorporated. (2011). 21. Lefebvre, A. H., Atomization and Sprays, Hemisphere Publishing,