1266
Energy & Fuels 2008, 22, 1266–1274
Diesel and Biodiesel Fuel Spray Simulations Primoz Pogorevc,* Breda Kegl, and Leopold Skerget UniVersity of Maribor, Faculty for Mechanical Engineering, SmetanoVa ulica 17, SI-2000 Maribor, SloVenia ReceiVed September 11, 2007. ReVised Manuscript ReceiVed October 28, 2007
This paper deals with the investigation of the influential parameters of a mathematical spray breakup model using different fuels. Beside injection system measurements, fuel physical properties and injection process characteristics were measured, because they are necessary for the spray simulation input. For validation purposes, spray was injected into motionless air at atmospheric pressure and room temperature and filmed with a highspeed camera. Spray macrocharacteristics have been determined on the recorded images. Using the simulation program, the injection processes for diesel, biodiesel, and their 50% blend B50 have been simulated. Spray mathematical model parameters were tuned based on the experimentally gained results. Primary breakup model parameters showed the biggest impact on the spray characteristics and were therefore expressed using the fuel physical properties, the injection process characteristics, and the working regime parameters. Spray simulations into the combustion chamber were made in the end. All of these results are presented and discussed in this paper.
1. Introduction Global atmosphere pollution has become a serious problem of today. The emissions from the combustion of fossil fuels contribute a notable part to this pollution. Environmental care together with the limited stock and growing prices of fossil fuels has given alternative fuels the potential to supplant a significant proportion of engine fuels. Application of the alternative fuels provides technical means to decrease emissions of conventional diesel engines. Biodiesel fuels, an alternative for petroleum fuel, are the acid esters of triglycerides that originate from vegetable or animal sources. This renewable source significantly reduces exhaust emissions of particulate matter, carbon monoxide, carbon dioxide, and hydrocarbons in comparison to diesel fuel.1–3 While biodiesel fuel has similar physical properties, no major engine modifications are necessary. Nevertheless, the differences in physical properties may result in spray anomalies causing an unnecessary increase of pollutive emissions. Higher viscosity affects the line pressure, leakage, and friction of the plunger in the pump, the rate of liquid pressure rise goes up, and the injection timing is advanced, because of the higher biodiesel bulk modulus. Higher injection pressure increases relative velocity between injected fuel and cylinder charge resulting in an average droplet size decrease, which improves fuel spray evaporation characteristics. A higher initial spray velocity also influences its penetration length and cone angle. On the other hand, the higher biodiesel density, viscosity, and surface tension increase the friction between the fuel and the nozzle wall and therefore decrease the injection velocity. Having * Corresponding author. E-mail:
[email protected]. (1) Graboski M. S.; McCormik R. L. Combustion of fat and vegetableoil derived fuels in diesel engines. Prog. Energy Combust. Sci. 1998, 24, 125–164. (2) Chang, D. Y. Z.; Van Grepen, J. H.; Lee, I.; Johnson, L. A.; Hammond, S. J.; Marley, S. J. Fuel properties and emissions of soybean oil esters as diesel fuel. JAOSC 1996, 73 (11), 1549–1555. (3) Choi, C. Y.; Bower, G. R.; Reitz, R. D. Effects of Biodiesel Blended Fuels and Multiple injections on D.I. Diesel Engine Emissions; SAE 970218, Society of Automotive Engineers: Warrendale, PA, 1997.
increased fuel viscosity and surface tension cause larger average droplet size and slower droplet vaporization. Fuels with a higher density and viscosity show shorter penetration lengths and greater spray cone angles; the mixing of air and fuel can be deteriorated, and a fuel-rich mixture is formed.4–8 Applicative research based on the fundamental knowledge of two-phase flows is used for the investigations of the fuel spray characteristics in individual industrial devices. They result in numerous empirical correlations, which are common in today’s practice and deliver important data for the development of spray mathematical models. Determination of the correlation of these empirical parameters presents difficulty, because they depend on the fuel injection system characteristics, working conditions, and fuel properties. Especially the primary breakup model, that takes place in the region close to the nozzle, is not only determined by the interaction between the liquid and gaseous phases but also by internal nozzle phenomena like turbulence and cavitation, and therefore, its parameters depend on fuel properties, nozzle geometry, and injection process characteristics. The breakup induced by turbulence and cavitation competes with the aerodynamic breakup until at a certain distance downstream of the nozzle exit the aerodynamic breakup processes become dominant. Higher injection velocities increase (4) Lee, S.; Tanaka, D.; Kusaka, J.; Daisho, Y. Effects of diesel fuel characteristics on spray and combustion in a diesel engine; JSAE200224660, Japanese Society of Automotive Engineers: Tokyo, Japan, 2002. (5) Lee, C. S.; Park, S. W.; Kwon, S. I. An Experimental Study on the Atomization and Combustion Characteristics of Biodiesel-Blended Fuels. Energy Fuels 2005, 19, 2201. (6) Wengiao, Y. Computational modelling of nitrogen oxide Emissions from biodiesel based on accurate fuel properties. W. Yuan, PhD Dissertation, University of Illinois at Urbana–Champaign. Department of Agricultural and Biological Engineering, 2006. (7) Allen C. A. W. Prediction of biodiesel fuel atomization characteristics based on measured properties. PhD Dissertation, Dalhousie University DALTECH, Halifax, Canada, 1998. (8) Yamane, K.; Ueta, A.; Shimamoto, Y. Influence of Physical and Chemical Properties of Biodiesel Fuel on Injection, Combustion and Exhaust Emission Characteristics in a DI-CI Engine, The Fifth International Symposium on Diagnostics and Modeling of Combustion in IC Engines COMODIA, Nagoya, Japan, July 1–4, 2001.
10.1021/ef700544r CCC: $40.75 2008 American Chemical Society Published on Web 01/12/2008
Fuel Spray Simulations
Energy & Fuels, Vol. 22, No. 2, 2008 1267
Table 1. Fuels Density, Viscosity, and Surface Tension at 30 °C fuel
density [kg/m3]
viscosity [Pa s]
surface tension [N/m]
D2 B50 B100
818.6 844.1 869.6
3.34 3.53 5.15
0.0259 0.0267 0.0276
turbulence levels which directly lead to higher turbulenceinduced breakup rates. The aerodynamic breakup rates are low at the nozzle exit, but they increase significantly moving away from the nozzle, where the compact liquid core has already been significantly disintegrated due to the primary breakup mechanisms. Achieving the accurate spray simulation results can lead to tedious tuning of mathematical model parameters.9 The aim of this paper is to illustrate the impact of the spray breakup mathematical model influential parameters on the spray characteristics using different fuels. The physical properties of diesel, biodiesel, and their blend B50, such as density, viscosity, surface tension, and speed of sound, were measured and used in the numerical simulations. Injection process characteristics such as injection pressure, nozzle needle lift, injection rate, and volume of injected fuel were recorded on the fuel injection systems test bench. Simultaneously, the spray penetration into motionless air at atmospheric pressure and room temperature was recorded using the high-speed camera. The spray simulations for both fuels and their blend B50 were made using AVL 3D program Fire v8.4. Breakup parameters were tuned for diesel and biodiesel fuel in order to achieve experimentally gained spray macrocharacteristics. With the use of appropriate expressions, Values of the most influential parameters of the mathematical breakup model were determined in dependence on the fuel’s physical properties, injection characteristics, and working regime. These expressions were tested on the diesel/biodiesel blend B50 at a different camshaft speed. Biodiesel and diesel spray simulations into the combustion chamber were made in the end. 2. Experimental Details For simulation input purposes, fuel physical properties and necessary injection process characteristics were measured. In order to validate numerical results, spray development was filmed using a high-speed camera. 2.1. Physical Properties. The physical properties of diesel, biodiesel, and their blend B50 were measured. Biodiesel used in our experiment was produced from rapeseed oil in the Pinus refinery,10 while the diesel fuel was supplied by the Petrol company.11 Their densities have been determined with a density meter DMA 35 PAAR, and viscosities were measured with the Herzog Ubbelohde Viscometer HVU 480, while the surface tension σ was calculated from the observed rise of a liquid in a thin capillary (eq 1). Some of the measured fuel properties are given in Table 1. 1 r σ ) Fgr h + 2 3
(
)
(1)
where F is the fuel density [kg/m3], g is the acceleration due to gravity [m/s2], h is the capillary rise [m], and r is the radius of the capillary [m]. 2.2. Injection process characteristics. Injection characteristics were measured on the fuel injection systems test bench Friedmann-
(9) Tatschl, R.; Künsberg Sarre, C.; Bergic, E. Engine spray modellingStatus and outlook, International Multidimensional Engine Modeling User’s Group Meeting at the SAE Congress, Vienna (Austria), July 7–12, 2002. (10) Pinus homepage. http://www.pinus-tki.si/ (accessed Nov 2007). (11) Petrol homepage. http://www.petrol.si (accessed Nov 2007).
Figure 1. Injection rate measurement device scheme.
Maier type 12 H 100-h.12 The conventional fuel injection system with a high pressure pump (Bosch PES6A95D410LS2542) and a one hole injection nozzle with a diameter of 0.68 mm (Bosch DLLA5S834) were used. A piezzo-resistive sensor (Kistler, type 6227) was used for the injection pressure measurements at the end of the high pressure pipe, while the nozzle’s needle lift was observed using the inductive sensor developed in our laboratory. Injection rate (IR) was determined using measured injection pressure (Figure 1), injection duration and the amount of injected fuel: pi,norm )
pi max(pi)
I ) √pi,norm∆φ
(2) (3)
Qc I (4) sum(I) IR ) MI∆φ (5) where pi is the measured pressure of the spray, max(pi) represents the maximal measured pressure of the spray, pi,norm is the normized spray pressure, Qc stands for the amount of the injected fuel, and φ represents the pump camshaft angle. All the data was stored on the computer using the program LabVIEW. Three different camshaft speeds (500, 800, and 1100 rpm) at full load were analyzed. Injection characteristics (injection pressure, needle lift, and injection rate) at 1100 rpm for diesel and biodiesel fuel are shown in Figures 2 and 3. The two diagrams show that the injection pressure curves are similar for both fuels, but maximal injection pressure is 55 bar higher in the case of biodiesel. Needle lift curves show that the injection is advanced when biodiesel is used. The differences are the consequence of the fuels’ physical properties and the speed of sound. 2.2. Spray Development. The fuel spray was injected into the glass chamber at room temperature and atmospheric pressure. For the spray filming, the high-speed digital camera Phantom v4.1 was used. The camera was placed at the distance of 2.5 m from the spray, about 20 cm below the nozzle hole. Due to the spray nature, the frame rate of 2500 fps at the resolution 128 × 512 pixels was chosen. The camera was triggered with falling electric pulse which was recorded together with the signal of TDC (top dead center), so that the time delay of the first recorded image could be defined. Diagrams in Figures 4 and 5 show needle lift and spray development filmed with a high-speed camera for diesel and biodiesel fuel at 500 rpm. Three consecutive spray developments for each fuel were filmed. The snapshot moments are marked below the images and on the needle lift curve. The differences in physical properties results in a steeper rise and higher injection pressure of biodiesel influencing the spray, which is narrower and about 5% longer at the end of the injection MI )
(12) Kegl, B. Numerical analysis of injection characteristics using biodiesel fuel. Fuel 2006, 85, 2377–2387.
1268 Energy & Fuels, Vol. 22, No. 2, 2008
PogoreVc et al.
Figure 2. Diesel injection process characteristics at 1100 rpm.
Figure 3. Biodiesel injection process characteristics at 1100 rpm.
(needle closing) than diesel. Because of relatively large nozzle diameter (0.68 mm), higher biodiesel viscosity shows less influence on the injection velocity and cone angle as the increase of the injection pressure.
3. Simulation The spray was simulated by the AVL 3D program Fire v8.4 using the Euler-Lagrangian approach. With respect to the liquid phase, spray calculations are based on a statistical method referred to as the discrete droplet method. Droplet parcels are introduced in the flow domain with initial conditions of position, size, velocity, temperature, and number of particles in the parcel. The droplets are tracked in the Lagrangian way through the computational grid used for solving the gas phase partial differential equations. Full two-way interaction between the gas and liquid phases is taken into account. The basic equation for the particle acceleration duip/dt is13 duip 3 Fa 1 Fa ) CD |u - uip|(uia - uip) + 1 g (6) dt 4 Fd Dd ia Fd i
( )
(13) AVL Fire Spray, version 8; AVL LIST Gmbh: Graz, 2005.
and can be integrated to get the particle velocity, and from this, the instantaneous particle position vector can be determined by integrating. In eq 6, Fa and Fd stand for air and fuel density, Dd is droplet diameter, and the uia - uip difference is between the particle and surrounding air velocities. Here, gi is the acceleration of gravity and CD stands for the drag coefficient expressed as follows:
{
24 (1 + 0.15Red0.687) CD ) Red 0.44
W Red < 103
(7)
W Red g 103
where Red is the particle Reynolds number. A cylindrical mesh with higher density in the middle section and at the nozzle area was used (Figure 6). The “diesel core injection” model was chosen to calculate the primary breakup of the spray, while the secondary breakup was simulated with the “wave” model. The primary breakup model considers two independent mechanisms: aerodynamic surface wave growth and internal stresses caused by the turbulence. The coherent liquid core region at the nozzle exit where primary breakup occurs is calculated from a mass balance of the liquid core at volume elements forming the core shape (Figure 6).
Fuel Spray Simulations
Energy & Fuels, Vol. 22, No. 2, 2008 1269
Figure 4. Diesel spray development at 500 rpm (three consecutive sprays).
Figure 5. Biodiesel spray development at 500 rpm (three consecutive sprays).
Figure 6. Mesh and primary breakup model used for the simulations.
Leading primary breakup equations which determine droplets initial conditions are the following:13 τA ) C1τT + C3τW
(8)
1.5
LA ) C2Cµ
k ε
(9)
Dd ) C2Cµ
k1.5 ε
(10)
k1.5 ε Vdrop ) C1τT + C3τW C2Cµ
(11)
where τA is breakup time (subscript T for the turbulent part and subscript W for the aerodynamic part), LA stands for turbulent length scale, Dd represents the initial droplet diameter, and νdrop is the droplets’ initial velocities. C1, C2, and C3 are primary breakup parameters and must be determined based on own experiences. They depend on injection system characteristics, fuel properties, the working regime, and other parameters; therefore, finding their right values and achieving accurate spray development can present a problem. In the wave secondary breakup model, two regimes are treated: one at high spray velocity conditions and one at low spray velocity conditions using the Rayleigh approach. In the first case, the initial size of the droplet diameter is supposed to be equal to the wavelength of the fastest growing or most probable unstable surface wave. The Rayleigh type breakup model produces droplets that are larger than the original parent drops. It is assumed that this regime does not have a significant effect on spray formation under conditions typical for highpressure injection.13 The influence of the mathematical spray model parameters was analyzed and tuned for diesel and
1270 Energy & Fuels, Vol. 22, No. 2, 2008
PogoreVc et al.
Figure 7. Spray development (diesel, 500 rpm).
Figure 8. Spray development (biodiesel, 500 rpm).
biodiesel fuel in order to achieve macrocharacteristics of the filmed spray. Fuel physical properties and injection process characteristics were used in the AVL Fire v8.4 “solver steering file” or as the “user” functions. Simulation results are shown in comparison with the filmed spray development at 500 rpm in Figures 7 and 8. In order to achieve good accordance in the spray shape and its development (the differences in the penetration lengths are less than 5%) between the numerical and experimental results, several trial runs with different breakup parameters were made. Because they depend on injection system characteristics, fuel properties, the working regime, and other parameters, despite spray simulation experiences, their determination can be very time-consuming. Experimental results are needed for their validation. Similar
results are shown at different high-pressure pump camshaft speeds, so identical findings can be drawn. Several trial runs pointed out that the most influential parameters on the spray macrocharacteristics (spray shape, penetration, and angle) are primary breakup parameters C1, C2, and C3. 4. Primary Breakup Coefficients Primary breakup parameter values were determined based on our experiences and several simulation runs in order to achieve actual spray characteristics. They proved to have a major impact on the spray. Their influence is shown in Figures 9 and 10. Arrows point out the value increase of parameters C1 (5, 12.5, 20) and C3 (0.1, 0.55, 1), while parameter C2 (0.2, 2) has a
Fuel Spray Simulations
Energy & Fuels, Vol. 22, No. 2, 2008 1271
Figure 9. Primary breakup parameter influence (C2 ) 2).
Figure 10. Primary breakup parameter influence (C2 ) 0.2). Table 2. Primary breakup parameters values C1/C2/C3 fuel diesel biodiesel
camshaft speed [rpm] 500 40/5/0.1 10/0.8/0.1
1100 10/0.5/0.6 1/0.2/0.2
lower value in Figure 10. All the other input data used was the same (diesel fuel at 500 rpm) for all simulations. The increase of parameter C1 (turbulent part of the breakup time) value increases spray penetration length, while the parameter C3 (aerodynamic part of the breakup time) determines the basic shape of the spray. The increase of the parameter C2, which directly influences the droplet initial diameters and velocities (eqs 10 and 11), generally decreases the spray angle and lengthens its penetration. Results of the tuned C1, C2, and C3 values for both fuels and camshaft speeds 500 and 1100 rpm are shown in Table 2. Parameters C1 and C3 (eq 8) determine the rate of aerodynamic and turbulent breakup. In the case of biodiesel fuel simulation, results indicate a considerably smaller turbulenceinduced breakup and a slight increase in the aerodynamic breakup rate. This is in accordance with differences in fuel physical properties, higher biodiesel viscosity, and surface tension resulting in bigger initial droplet diameters. Parameter C2 influences the turbulent length scale and droplet initial velocities (eqs 10 and 11), affecting the cone angle, which are smaller when biodiesel fuel is used. That deviates from general assumptions and is probably the consequence of the large nozzle diameter. Several different equation formations for the primary breakup parameters were tested. The best results were gained if the following expression was used: Ci ) xa11xa22xa33...
(12)
where χ represents fuel physical properties, injection characteristics, and working regime parameters and ai are appropriate coefficients. Coefficients a1, a2, ... in the nonlinear equation system were determined with the Newton–Raphson method. Necessary
derivatives were calculated with the package AceGen in the program Mathematica.14 The influence of fuel physical properties, injection characteristics, and working regime parameters on each primary breakup parameter was analyzed. Numerous trials gave the best results for the pretuned C1, C2, and C3 values using following parameters and appropriate coefficients in the following expressions: C1 ) Ff7.808µf-2.276σf14.676tinj-31.348sqn-8.391
(13)
C2 ) Ff-11.507µf-0.912σf-23.985Qc38.725pave15.9889n-12.420 (14) C3 ) Ff13.447µf2.369σf25.764tinj-69.983pave-23.873n-4.015
(15)
where Ff is fuel density µf is fuel viscosity [mPa s], σf is fuel surface tension [N/mm], tinj stands for injection time [ms], pave is average injection pressure [MPa], sq ) pave/pmax (squarness), Qc represents fueling [mm3/cycle], and n is pump speed [1/min]. Fuel density, viscosity, and surface tension together with pump speed appear in expressions for all three parameters. The average and maximal injection pressure ratio seem to influence turbulent induced breakup, while average pressure determines the aerodynamic breakup rate. If we check coefficients in the expressions, fuel density and surface tension are the most influential for spray breakup rate parameters. Injection duration affects them both; however, its influence is not very significant. Injected fuel quantity and average injection pressure, which directly influence on the initial injection velocity and droplets diameters, appear in the expression for the parameter C2. That is in accordance with eqs 8 and 9 of the mathematical primary breakup model. These expressions for the primary breakup model parameters were inserted into the AVL Fire v8.4 solver steering file and tested on the blend of diesel and biodiesel B50 fuel at 800 rpm. Therefore, blend physical properties and operating parameters determined the necessary parameters, which were used in the simulation. Recorded spray development in comparison with [kg/m3],
(14) Korelc, J. Automatic generation of numerical codes with introduction to AceGen 4.0 symbolic code generator. University of Ljubljana, Faculty of Civil and Geodetic Engineering, www.fgg.uni-lj.si/Symech/, 2005.
1272 Energy & Fuels, Vol. 22, No. 2, 2008
PogoreVc et al.
Figure 11. Spray development (B50, 800 rpm).
Figure 12. Piston and spray direction.
simulation results is shown in Figure 11. Comparison shows good accordance in the spray development as well as in its shape (spray angle and penetration). The difference between numerical and actual penetration is less than 5%. Spray simulation results for B50 confirm the adequacy of primary breakup parameter expressions for a used conventional injection system. In order to expand the application area, the injection system geometry parameters should be considered in the expressions. 5. Combustion Chamber Spray Simulations Spray simulations into the combustion chamber of bus engine MAN D2566 were carried out subsequently. Real pressure, temperature, and air swirl were considered. The combustion chamber (piston) together with the spray direction are shown in Figure 12. The cylinder head, which closes the chamber on top, is flat. Measured fuel physical properties, injection characteristics, and tuned spray model parameters were used in the
simulations. The spray development of diesel and biodiesel is shown in section A-A of the combustion chamber (Figure 13). They are shown as if the start of the injection for both fuels was the same. Differences between diesel and biodiesel spray are noticeable in Figure 14. They appear to be smaller as in shown in the spray simulations into the air at room temperature. Higher velocities in the middle of the spray, which indicate a higher fuel concentration, confirm the worse atomization and slightly longer penetration in the case of biodiesel fuel, while the cone angle appears to be almost the same. MAN D2566 has M system injection, so fuel distribution on piston walls is very important. Figures 14 and 15 show velocities on the piston wall from the side and on the interesting halfslooking from the bottom up. Higher velocities areas signify fuel presence and can be used for spray distribution evaluation. Larger fuel-rich areas on piston walls of biodiesel spray in comparison with diesel are more noticeable at 1100 rpm (Figure 15). It can be said that, in the case of the unmodified conventional injection system with a relatively large nozzle diameter, the biodiesel physical properties increase the injection pressure resulting in higher spray velocities and longer penetration. Denser biodiesel spray therefore distributes over a larger piston wall area, which could mean positive influence of biodiesel physical properties. On the other hand, simulation shows that biodiesel droplets are around 20% bigger, and therefore, their evaporation process is longer; so, undesired fuel-rich mixture areas can occur. 6. Conclusions The influence of fuel on the spray characteristics were investigated in this paper. Spray development was recorded with
Fuel Spray Simulations
Energy & Fuels, Vol. 22, No. 2, 2008 1273
Figure 13. Spray development shown on section A-A.
Figure 14. Spray distribution at 500 rpm.
the high-speed camera for diesel, biodiesel, and their blend B50 at different pump speeds. The spray simulations for both fuels were performed into motionless air at room temperature and atmospheric pressure in order to tune the mathematical model parameters. Primary breakup parameters showed the most influence on the spray characteristics and were therefore
expressed with fuel physical properties and operating regime parameters. Real condition spray simulations were performed in the end. The results gained lead to the following conclusions: (1) The difference in the fuel’s physical properties is reflected in the injection characteristics of the unmodified conventional fuel injection system. A higher biodiesel density and bulk
1274 Energy & Fuels, Vol. 22, No. 2, 2008
PogoreVc et al.
Figure 15. Spray distribution at 1100 rpm.
modulus result in an injection pressure increase in the conventional injection system, which is reflected in advanced injection, higher spray velocity, narrower spray, and larger penetration lengths. The higher fuel viscosity and surface tension, which generally decrease the initial spray velocity and widen the cone angle due to an increase in friction, do not have significant influence because of the large nozzle diameter. However for spray simulation purposes, just changing the fuel properties and injection process characteristics is not always enough and mathematical model influential parameters need tuning. (2) Primary breakup parameters have a huge impact on the spray macrocharacteristics (shape, angle, and penetration length of the spray). There is no exact formula for their values, and they must be determined based on experiences or tuned based on the experimental results. The paper shows that with the fuel physical properties, operating conditions, and injection characteristics, the physical-mathematical model’s empirical parameters can be expressed. With use of the appropriate expressions, empirical parameters values can be determined and the fuel spray characteristic for different kinds of fuels can be predicted. It has to be pointed out that the presented expressions suit only
the conventional injection system used in our research, which is very specific (nozzle with only one hole of relatively large diameter). But nevertheless, this example shows that these parameters can be expressed in the way that spray simulations for different fuels and working regimes can be made. In order to get more universal expressions, similar tests would have to be made for several different injection systems. Their specifications, such as nozzle hole diameter, number of holes, and others should be included in the expressions. (3) In the unmodified conventional injection M system, biodiesel physical properties showed positive influence by distributing spray over a larger area of the piston chamber. Nevertheless, there are also some negative factors, such as bigger spray droplets and a slower evaporation process, which must be taken into consideration. Acknowledgment. This research was supported by the European Community’s Sixth Framework Programme in the scope of the Civitas II Mobilis project. EF700544R
