Numerical and Experimental Analysis of ... - ACS Publications

1418

Energy & Fuels 2008, 22, 1418–1424

Numerical and Experimental Analysis of Combustion and Exhaust Emissions in a Dual-Fuel Diesel/Natural Gas Engine† Stefano Cordiner,‡ Michele Gambino,*,§ Sabato Iannaccone,§ Vittorio Rocco,‡ and Riccardo Scarcelli‡ Department of Mechanical Engineering, UTV, UniVersity of Rome Tor Vergata, Rome, Italy, and IM-CNR, Istituto Motori of Italian National Research Council, Naples, Italy ReceiVed August 4, 2007. ReVised Manuscript ReceiVed January 3, 2008

The dual-fuel diesel/natural gas (NG) concept is based on the introduction of a homogeneous charge of air and NG in a diesel engine. The charge is then ignited by the combustion of a small amount of diesel fuel directly injected into the cylinder near the top dead center. Such technology can be considered as a promising and inviting solution to address the tradeoff of particulate matter-NOx emissions affecting traditional diesel engines. In this paper the conversion of a heavy-duty diesel engine to dual-fuel operations is discussed. Experimental tests were performed to define engine performance and reduce exhaust emissions. The experimental data also were used to develop a numerical analysis, characterized by a mixed one-dimensional (1D)-threedimensional (3D) approach. As far as 3D simulation tools are concerned, a modified version of the KIVA-3V code was used to simulate the whole working cycle of the engine and to represent diesel injection and overall combustion.

Introduction Dual-fuel engines are one of the possible short-term solutions to reduce emissions from traditional diesel engines (affected by the problem of the tradeoff of particulate matter (PM)-NOx emissions) while utilizing an alternative fuel like natural gas (NG). It consequently results in an interesting technology to meet future emissions regulations and a powerful solution to retrofit existing and running on-road diesel engines, thus shortening the period required for a new regulation to be effective on the air quality. In a dual-fuel engine the primary fuel (NG) is mixed with air into the intake manifold, just like a spark ignition (SI) engine. The mixture is then ignited by the combustion of a small amount of diesel fuel (the pilot) injected as the piston approaches the top dead center (TDC). The wide zone of ignition (as a result of the presence of a large number of flame kernels in the chamber instead of a localized spark plug) leads to a regular combustion of the whole charge into the cylinder also for ultralean mixtures, thus extending the operating flammability limits with respect to SI-NG operations. The high octane number (NOR ) 130) of NG also prevents the occurrence of knocking. Accordingly, the use of NG as primary fuel allows the same compression ratio of the conventional diesel engine; thus, existing diesel engines can be easily converted to dualfuel operation. In fact, the only pieces of equipment that have to be installed on the engine are the feeding system for NG and the external devices allowing variation of the diesel injection flow rate. This results in great savings in capital cost and time in comparison with other environmental low-impact solutions (fuel cells, hybrid vehicles). †

Presented at the 10th International Conference on Energy and Environment. * To whom correspondence should be addressed. Telephone: +39.081. 7177140. Fax: +39 0812396097. E-mail: [email protected]. ‡ University of Rome Tor Vergata. § Istituto Motori of Italian National Research Council.

Dual-fuel pilot-injected engines have a significant potential to reduce NOx and PM emissions. For high percentages of NG utilization, the diesel fuel can only be considered as the ignition source for NG combustion, with consequent benefit on PM emissions. Moreover, the combustion of a homogeneous charge should provide reduced NOx with respect to the baseline diesel case. Also, good values of the fuel specific consumption and the low atomic ratio C/H of NG (mostly composed by methane) allow the reduction of carbon dioxide (CO2) emissions, which is not yet ruled as but is considered as one of the most important causes of the greenhouse effect. In recent years, the dual-fuel combustion system has been proposed and studied1–5 but has not reached a great diffusion. This can be explained considering that HC and CO emissions from a dual-fuel engine are much higher than those of a conventional diesel engine, especially at part loads. Dual-fuel engines, deriving from the conversion of conventional diesel engines, usually operate unthrottled, with the load regulated by the admission of NG in the intake manifold. The homogeneous charge becomes leaner as the load is reduced, thus exceeding the lean flammability limit. Combustion becomes irregular and incomplete, so leading to the increase of CO and HC in the exhaust emissions. Other sources for HC emissions are the scavenging phase and the crevice volumes. Unlike a conventional diesel engine, during dual-fuel operations, the charge entering in the cylinder through the inlet valve is a mixture of air and NG. Thus, a fraction of NG is forced into crevice volumes during the compression stroke or escapes through the exhaust valve during the valve overlap period (short circuit effect). The latter aspect is particularly relevant when the overlap (1) Weaver, C. S.; Turner, S. H. SAE Paper 940548, 1994. (2) Karim, G. A.; Liu, Z.; Jones, W. SAE Paper 932822, 1993. (3) Ishida, A.; Nishimura, A.; Uranishi, M.; Kihara, R.; Nakamura, A.; Newmann, P. JSAE ReView 2001, 22, 237–243. (4) Liu, Z.; Karim, G. A. SAE Paper 950466, 1995. (5) Papagiannakis, R. G.; Hountalas, D. T. Int. J. Appl. Therm. Eng. 2003, 23, 353–365.

10.1021/ef7004755 CCC: $40.75  2008 American Chemical Society Published on Web 02/21/2008

Analysis of Combustion and Exhaust Emissions

is great, as is usual for turbo-charged diesel engine. The low efficiency and the high emissions under part loads represent, together with the possible occurrence of knocking at high loads (unavoidable for mixtures close to stoichiometric conditions), the major problems for a dual-fuel conversion of traditional diesel engines. Many possible solutions have been considered to improve the behavior of dual-fuel engines under part load. A possible approach is the optimization of the pilot injection characteristics: When the pilot quantity is increased and the injection timing is advanced, a more regular combustion process, with higher efficiency and lower HC and CO emissions, can be observed.6,7 Nevertheless, these solutions can be responsible for an increase in PM and NOx emissions. Another approach is to reduce the amount of air in the cylinder by introducing intake throttling or varying intake valve timing. Also these solutions can give high PM, CO, and NOx emissions. Recent studies8 show that the use of low pilot quantities (about 2–3%), together with advanced pilot injection (45–60 CAD BTDC), permits one to obtain very low NOx emissions and “diesel-like” performance. Daisho et al.9,10 have carried out engine tests modifying some engine parameters such as pilot injection timing advance, intake throttling, and hot and cooled exhaust gas recirculation (EGR). It was found that hot EGR is the best way to increase thermal efficiency and reduce CO/HC emissions at light loads while maintaining low levels of NOx. With advanced injection timing, a better thermal efficiency can be obtained; however, the resulting NOx emissions are higher. Intake throttling promotes better combustion notwithstanding the increasing pumping losses. Even though many studies were carried out in the recent years, the dual-fuel combustion process is still unclear. Moreover, the present experimental analysis, although it demonstrated the dual-fuel benefits, has not allowed a complete tailoring of the engine to explore all the potential of this technology. To this end, a numerical analysis, with the goal to correctly represent dual-fuel combustion and afterward to guide the experimental procedure, has been undertaken. Dual-fuel combustion depends highly on local fluid dynamics. As a consequence, a complete simulation of the engine is required and has been performed by using a mixed one-dimensional (1D)-three-dimensional (3D) approach. 3D tools have been mainly used to model the in-cylinder thermo-fluid dynamic processes; 1D tools have been used to compute realistic timedependent 3D-code boundary conditions by means of which the presence of the overall engine ducts and devices can be taken into account. 1D Tools. The integrated 0D-1D code FW2001 has been entirely developed in recent years at the University of Rome Tor Vergata. It has been successfully used to predict SI engine performance with regard to simple combustion chamber geometries. When complicated geometries are concerned, a more detailed (3D) approach is desirable to take into proper account nonaxial-symmetric chambers. Nevertheless, FW2001 can be (6) Abd Alla, G. H.; Soliman, H. A.; Badr, O. A.; Abd Rabbo, M. F. Energy ConVers. Manage. 2000, 41, 559–572. (7) Abd Alla, G. H.; Soliman, H. A.; Badr, O. A.; Abd Rabbo, M. F. Energy ConVers. Manage. 2002, 43, 269–277. (8) Srinivasan, K. K.; Krishnan, S. R.; Singh, S.; Midkiff, K. C.; Bell, S. R.; Gong, W.; Fiveland, S.; Willi, M. Proceedings of the International Joint Power Generation Conference (IJPGC), Paper 40098; Atlanta, GA, June 16–19, 2003. (9) Daisho, Y.; Yaeo, T.; Koseki, T.; Saito, T.; Kihara, R.; Quiros, E. SAE Paper 950465, 1995. (10) Daisho, Y.; Takahashi, Y. I.; Nakayama, S.; Kihara, R.; Saito, T. SAE Paper 952436, 1995.

Energy & Fuels, Vol. 22, No. 3, 2008 1419

successfully used to simulate the presence of the overall engine ducts and devices to provide, starting from the experimental pressure trace, the actual and time dependent boundary conditions to the 3D code, which in turn computes the 3D chamber and intake-exhaust duct fluid dynamics. Details on the engine schematization and the 1D approach can be found in a previous paper by the authors.11 3D Tools. The multidimensional code utilized in this work is a modified version of the KIVA-3V code.12 It has been used to successfully model gasoline, diesel, and NG fueled engines since its inception, and it still represents the best code to simulate in cylinder processes. Nevertheless, some modifications are needed to improve its predictive capabilities with special regard to the representation of spray, ignition, and combustion processes. Spray Model. The present work utilizes the well-known WAVE model, developed by Reitz and Bracco,13,14 to represent both the primary breakup (liquid atomization) and the secondary breakup (drop breakup) processes. The main assumption of the WAVE model is that the atomization of the injected liquid is the result of the aerodynamic interaction between the liquid jet and the surrounding gas. Such interaction leads to the propagation of unstable waves on the liquid surface. The WAVE model computes breakup by using results from a stability analysis. The breakup time is equal to 3.788B1a (1) ΩΛ where B1 is the breakup time constant, a is the parent drop or jet radius, Ω is the wave fastest growth rate, and Λ is its wavelength. Reitz and Diwakar15 applied the “wave” atomization theory to drop breakup modeling. The injection of a liquid jet is simulated through the introduction of a sequence of liquid discrete parcels (blobs) having a size calculated as a distribution about the sauter mean radius of the droplets at the injector. The mass of new droplets formed as a result of breakup is subtracted from that of the parent drop. The change of the parent drop radius is τ)

(a - r) da )(2) dt τ The WAVE model assumes that the new droplets generated from the parent drop have a radius r equal to

{

r ) B0Λ

{

r ) min

if (B0A e a) 2

0.33

(3πa U/2Ω) (3a2Λ/4)0.33

if (B0A > a, only one time)

(3)

The recommended value for B0 is set equal to 0.61 for diesel sprays, whereas the breakup time constant B1 is related to the initial disturbance level in the breakup process and to the wall impingement effects. The suggested values are B1 ) 10-60 when low turbulence and low wall impingement are expected and B1 < 10 vice versa. In the present work, the authors fixed B1 ) 5 to correctly represent the interaction between the spray and the walls of the piston bowl. In addition to the aerodynamic interaction, other possible sources of liquid atomization are cavitation (Acroumanis model) and turbulent flow (Huh and Gosman model) in the nozzle. The present study does not (11) Cordiner, S.; Gambino, M.; Iannaccone, S.; Rocco, V.; Scarcelli, R.; Setaro, G. SAE Paper 2005-24-032, 2005. (12) Amsden, A. A.; O’Rourke, P. J.; Butler, H. E. KIVA III: A KIVA Program with Block Structured Mesh for Complex Geometries, LA-12503MS; Los Alamos National Laboratory: Los Alamos, NM, 1993. (13) Reitz, R. D.; Bracco, F. V. Phys. Fluids 1982, 25, 1730–1742. (14) Reitz, R. D. Atomization Spray Technol. 1987, 3, 309–337. (15) Reitz, R. D.; Diwakar, R. SAE Paper 870598, 1987.

1420 Energy & Fuels, Vol. 22, No. 3, 2008

Cordiner et al.

consider such effects because they are particularly important for high speed breakup (high injection pressure like in modern common rail systems) while the examined engine is characterized by a mechanical diesel injection (the injection pressure is about 240 bar). For the same reason, the contribution of drop distortion to the breakup process was also neglected, and the authors decided to not completely test other available breakup models like TAB16 (the original breakup model implemented in KIVA-3V) and DDB (a modified version of TAB developed by Ibrahim et al.17). n-Heptane (C7H16) was utilized as the vaporizing species representing the diesel fuel. Shell Ignition Model. The Shell model, developed by Halstead et al.,18 was originally drawn up to describe the knock phenomenon in gasoline engines, even though it subsequently was successfully used to simulate ignition in diesel engines.19,20 It describes the autoignition of hydrocarbon fuels by means of a simplified reaction mechanism characterized by eight singlestep reactions involving five generic species: Initiation: RH + O2 f 2R*

kinetic coeff. Kq

(4)

Propagation: R* f R* + P + Heat

kinetic coeff. Kp

R* f R* + Q

kinetic coeff. f4Kp

(7) (8)

Branching: kinetic coeff. Kb

(9)

Termination: R* f inert 2R* f inert

kinetic coeff. f3Kp

(10)

kinetic coeff. Kt

(11)

The term RH represents the hydrocarbon fuel (CnH2m), R/ is the radical formed from the fuel, B is the branching agent, Q is an intermediate species, and P represents the oxidized products (CO, CO2, and H2O). The rate constants in eqs 411 are defined as follows: fi ) Afi exp(-Efi/RT)[O2]xi[RH]yi

(12)

Kj ) Aj exp(-Ej/RT)

(13)

Kp )

[

1 1 1 + + kp1[O2] kp2 kp3[RH]

]

where Ym/ is the local and instantaneous thermodynamic equilibrium value of Ym. τc is the characteristic time, given by

where τl and τt are the laminar and the turbulent time scales, respectively. The coefficient f accounts for the influence of turbulence on combustion and works as a switch between laminar and turbulent combustion after ignition. The laminar time scale is calculated from the one-step reaction rate and is equal to

(6)

B f 2R*

(15)

τc ) τl + fτt

kinetic coeff. f1Kp kinetic coeff. f2Kp

Ym - Y m/ dYm )dt τc

(5)

R* f R* + B R* + Q f R* + B

1000-1100 K. Once the temperature exceeds such values, the Shell model has to be switched off. Within the present paper, the switch temperature was set at 1000 K. The Shell model simulates diesel autoignition; thus, only n-heptane was considered as fuel. Once the diesel fuel ignited, a proper combustion model has to be introduced. A possible approach is to extend the validity of a typical diesel combustion model (e.g., characteristic-time combustion, CTC, model) to dualfuel operations. When a low amount of diesel oil is utilized (high NG percentage), also a flame propagation model (e.g., coherent flame model, CFM) can be used to describe NG combustion only, starting from the ignition zones. CTC Model. The CTC model was originally introduced for SI engine modeling22 and subsequently extended to study diesel combustion.22 It involves seven species (fuel, O2, N2, CO2, CO, H2, and H2O) and assumes that the rate of change of the mass fraction Ym of the generic species m is given by

τl ) A-1[C7H16]0.75[O2]-1.5 exp(E/RT)

where j stands for p1, p2, p3, q, b, and t. Details on the Shell implementation and the utilized values for Afi, Efi, xi, yi, Aj, and Ej can be found in previous papers.18,20 The ignition process was assumed to occur in the local diesel/air equivalence ratio range between 0.5 and 3.21 The Shell model can be considered as a reliable tool to describe only the lowtemperature chemistry of hydrocarbon fuels, and its validity range usually goes up to temperature values close to (16) O’Rourke, P. J.; Amsden, A. A. SAE Paper 870289, 1987. (17) Ibrahim, E. A.; Yang, H. Q.; Przekwas, A. J. J. Propulsion 1993, 9, 651–654. (18) Halstead, M.; Kirsh, L.; Quinn, C. Combust. Flame 1977, 30, 45– 60. (19) Kong, S.-C.; Reitz, R. D. J. Eng. Gas Turbines Power 1993, 115, 781–789. (ASME Trans.) (20) Kong, S. C.; Han, Z.; Reitz, R. D. SAE Paper 950278, 1995. (21) Sazhina, E. M.; Sazhin, S. S.; Heikal, M. R.; Marooney, C. J. Fuel 1999, 78, 389–401.

(17)

where A ) 2.868 × 109 (SI units) is the pre-exponential factor, and E ) 125.6 kJ/mol is the activation energy. The turbulent time scale is modeled as k (18) ε where the standard k-ε model is used to describe turbulence into the cylinder and to calculate k and ε, and C2 is the CTC model constant. The switch factor f varies from 0 to 1 according with the ratio of combustion products r as follows: τt ) C2

1 - e-r 0.632 YCO2 + YH2O + YCO + YH2 f )

-1

(14)

(16)

r)

1 - YN2

(19) (20)

Thus, the early combustion lies in the laminar regime while turbulence becomes more important later on when combustion products are not negligible. In this paper, the CTC model was modified to account for dual-fuel combustion. Both fuels (nheptane and NG) are considered when computing the equilibrium mass fraction of each species and are burnt according to their proportions. Also, the respective values of the lower heating value are considered for diesel and NG combustion. CFM. The CFM was first introduced by Marble and Broadwell23 and later extended.24,25 Its formulation is based on the (22) Abraham, J.; Bracco, F. V.; Reitz, R. D. Combust. Flame 1985, 60, 309–322. (23) Marble, F.; Broadwell, J. Technical Report TRW-9-PU Project Squid; Purdue University West: Lafayette, 1977. (24) Candel, S. M.; Poinsot, T. J. Combust. Sci. Technol. 1990, 70, 1– 15.



hypothesis of flamelet combustion. The generalized flamelet assumption requires that chemical reactions take place in thin sheets that propagate at the laminar flame speed. The chemical reaction of fuel oxidation is supposed to occur in this very thin layer separating the burned and unburned gases. Although the flame thickness increases once the mixture is in a certain lean range, the role of chemical kinetics is more important. By examining flames at different equivalence ratios, it was shown that combustion can still occur in laminar flamelets. This implies that the chemical time scale is short in comparison with the turbulent time scale; therefore, the chemistry can be decoupled from the turbulence. The mean reaction rate then can be expressed by ω ) FuYF,uSLΣ

(21)

In the mean reaction rate model, two unknown parameters appear: the flame laminar speed SL and the flame surface per unit volume Σ. The former depends strictly on local thermochemical variables. In this paper it is calculated through the following equations:26 SL(Tu, p, φ) ) SL,0

( )( ) Tu T0

Rk

p p0

βk

(22)

where SL,0 is the unscratched laminar burning velocity at room temperature and pressure expressed as a function of the equivalence ratio as SL0(φ) ) akφRk exp-ck(φ - dk)2 

(23)

and where ak, bk, ck, dk, Rk, and βk are constants depending on the fuel. The latter term Σ can be described by an exact transport equation. In a turbulent flow field, the Σ equation becomes ∂Σ¯ + ∇·ν¯Σ¯ + ∇·ν′Σ¯ + ∇·(SLn)Σ¯ ) ksΣ¯ ∂t

(24)

where kS is the turbulent flame stretch. Substituting the CFM modeling assumptions, the final equation gives

(

)

βFuSLΣ¯2 νt ∂Σ¯ + ∇·ν¯Σ¯ ) aeΣ¯ + ∇· ∇·Σ¯ ∂t FYF σΣ

Table 1. Engine Specifications parameter

specification

engine type number of cylinders stroke bore volume displacement compression ratio maximum power maximum torque diesel injection system turbo-charging maximum TC pressure maximum IMEP

IVECO 8360.46R 4 stroke CI 6 inline 130 mm 112 mm 7.8 L 17.6:1 166 kW @ 2050 rpm 965 Nm @ 1250 rpm mechanical direct (240 bar) yes 1.3 bar @ 2050 rpm 15.8 bar @ 2050 rpm

Table 2. NG Composition and Properties component

volumetric concentration

methane ethane ethylene propane propylene acetylene propine i-butane n-butane i-pentane n-pentane benzene toluene xylenes aliphatic C6-C8 CO CO2 N2 stoich. air/fuel ratio lower heating value density

96.7 0.11 1.2 0.20 2.2 3.1 9.6 0.20 0.30 650 720 60 19 10 0.11 7 0.16 1.10 17.1 49.0 0.74

vol % vol % ppm vol % ppm ppm ppm vol % vol % ppm ppm ppm ppm ppm vol % ppm vol % vol % kg/kg MJ/kg kg/Nm3

introduced a G-equation model to simulate NG combustion.30 In this paper, the CTC model was utilized to describe dual-fuel combustion. Moreover, a comparison between the CTC model and the CFM was carried out to describe their potentialities and limitations.

(25)

where e represents the mean strain rate (equal to ceε/k, ce being a constant equal to 5) and R and β are the model constants. The CFM has been applied successfully27,28 to model combustion by means of the flame propagation mechanism and has provided significant results in terms of reliability of the predictions and accuracy of the results. Basic 3D analyses of dual-fuel combustion are reported in the literature. Reitz et al. 29 used the KIVA-3V code to simulate diesel ignition by means of the Shell model and dual-fuel combustion by the CTC model, properly modified to take into account NG as fuel in addition to diesel oil. They realized that, when the amount of NG increases (i.e., NG g 90% of the total energy available), the CTC is not able to describe the flame propagation throughout the charge. Therefore, they subsequently (25) Meneveau, C.; Poinsot, T. J. Combust. Flame 1991, 81, 311–332. (26) Abu-Orf, G. M.; Cant, R. S. Combust. Flame 2000, 122, 233–252. (27) Cheng, W. K.; Diringer, J. A. SAE Paper 910268, 1991. (28) Zhao, X.; Matthews, R. D.; Elley, J. L. SAE Paper 932713, 1993. (29) Singh, S.; Kong, S. C.; Reitz, R. D.; Krishnan, S. R.; Midkiff, K. C. SAE Paper 2004-01-0092, 2004.. (30) Singh, S.; Liang, L.; Kong, S. C.; Reitz, R. D. Int. J. Engine Res. 2006, 7, 65–75.

Experimental Section The experimental activity was carried out at IM-CNR. Details on experimental procedures and facilities can be found in previous papers by the authors.11,31 The main characteristics of the engine are listed in Table 1. First, experimental tests on dual-fuel mode showed high CO and HC emissions, especially under part load conditions. The IVECO 8360.46R engine is characterized by a mechanical diesel injection system, and the optimization of the pilot injection characteristics is only partially feasible. In fact, it is possible to modify the pilot quantity, whereas the pilot injection advance cannot be varied while the engine is working. Thus, the only possible solution to reduce exhaust emissions was to control the admission of air and NG into the cylinder. That is why intake throttling and EGR were then introduced. Moreover, a three-way catalyst was utilized to oxidize the residual HC and CO emissions. Cylinder pressure data were averaged over 256 engine cycles to account for cycle-to-cycle variations, and the resulting medium pressure cycle was utilized for numerical applications. Three repetitions for each measurement were carried out. The properties of NG used for the experimental test are reported in Table 2. The (31) Cordiner, S.; Gambino, M.; Iannaccone, S.; Rocco, V.; Scarcelli, R. SAE Paper 2007-24-0124, 2007.


Cordiner et al.

Table 3. 13-Mode Test Emissions

Table 4. Examined Test Cases

PM, BSEC, HC, CO, NOx, g/(kW h) g/(kW h) g/(kW h) g/(kW h) MJ/(kW h) European R49 limits (EURO2) full diesel dual-fuel 1 dual-fuel 2 dual-fuel average

1.1

4.0

7.0

0.15

0.6 0.5 0.3 0.4

2.2 2.8 0.1 1.5

6.5 1.2 3.9 1.6

0.26 0.07 0.07 0.07

9.8 11.1 11.1 11.1

lower heating value of diesel oil used is 42.6 MJ/kg versus 49.0 MJ/kg of NG.

Results and Discussion Experimental Results. During the experiments, a poor combustion quality in the dual-fuel operation was detected at low engine load, both at low speed (1250 rpm) and at high speed (1950 rpm). Therefore, in these conditions as for idle speed, the engine was run in full diesel operation. Instead at middle and high load, apart for maximum torque, the engine operation was dual-fuel throttled with EGR. The throttle was necessary to reduce oxygen excess (typical of a diesel engine) and to obtain a good performance from the three-way catalyst. The EGR was added to improve the diesel pilot ignition when the intake air quantity was reduced by throttling. In fact in these conditions inlet pressure is proportionally reduced, and therefore a bad combustion ignition quality can occur, for the consequently low pressure and temperature in the combustion chamber during the compression stroke. The EGR quantity at each point was chosen with the aim to obtain a good compromise between the necessity to “fell” the cylinder with an inert gas mixture and the necessity to have the right air quantity for a good flame front propagation without quenching to avoid high pollutant production (especially HC, CO, and PM). Furthermore, the EGR, together with the throttle valve, allowed us to reduce the knocking probability and therefore to operate with more safety at high load. In dualfuel operation, the engine was run unthrottled and without EGR only at maximum torque. Moreover, in this condition all the regulated emissions obtained, including PM emissions, were very low. On the basis of the above considerations two 13mode cycles (ECE R49) were performed in the following way: the points 1, 2, 3, 7, 11, 12, and 13 were performed in full diesel operation (1, 7, and 13 are at idle speed); the points 4, 5, 8, 9, and 10 were performed as throttled dual-fuel operations with EGR. Only point 6 (maximum torque) was performed as an unthrottled dual-fuel operation without EGR. The main remark regards the relatively high tendency to knock at higher loads, especially at maximum power, in spite of the EGR and the intake air reduction. This could be an issue for further tests, namely, to evaluate the possibility to control knocking better in dual-fuel operations or to operate the engine in full diesel, also at higher loads and rated power. Obviously this last opportunity has to be assessed considering the corresponding emission levels (especially NOx and PM) obtainable with the diesel operation. In Table 3, the 13-mode cycle emissions for the dual-fuel engine IVECO 8360-ETRA are reported. The results were obtained running some points in full diesel operation and others in dual-fuel operation, according the above description. The test cycle was performed two times, and the results were compared with those of a traditional EURO 2 diesel engine. Considerable differences can be appreciated, especially between the CO and the NOx emissions of the two tests. This is due to the presence of the catalyst. In fact a small variation in air fuel control (λ) can significantly affect the catalyst conversion efficiency. The average emission values are also reported in the

test case

MF3

MF6

engine speed load intake throttle EGR three-way catalyst diesel flow rate NG flow rate NG-air λ % NG (on energy basis) injection timing advance injection duration

1250 rpm 480 Nm WOTa no no 4.6 kg/h 9.0 kg/h 2.48 70 350 CAD (10 BTDC) 11 CAD

1250 rpm 480 Nm WOTa no no 1.64 kg/h 12.1 kg/h 1.84 90 350 CAD (10 BTDC) 4 CAD

a

WOT: wide open throttle.

Figure 1. Effect of the swirl motion on the evolution of the spray (MF3, CAD 360).

same table. The merits of the strategy adopted for the dual-fuel operation is evidenced by the large global emissions reduction in comparison with the diesel version. In particular, the NOx and PM levels obtained were very low and far from the EURO 2 limits while those of the diesel engine were much higher and close to these limits. As expected, lower differences in comparison with the diesel were found for HC and CO. Also for these two pollutants, a reduction was obtained with the dualfuel operation. In the same table, the Brake Specific Energy Consumption (BSEC) is also reported. The dual-fuel engine efficiency was slightly lower than the diesel (about 10%). This result is easily explained with throttled operation and slightly lower combustion efficiency in dual-fuel operation. Maximum torque and power in the dual-fuel mode resulted in slightly lower than full diesel (780 Nm vs 840 Nm both at 1250 rpm and 143 kW at 1950 rpm vs 154 kW at 2050 rpm, respectively). This small differences can be attributed to the modifications of the following: (1) the intake manifold with a throttle valve, (2) the exhaust pipe with EGR and a catalyst, (3) and the diesel ignition system. In dual-fuel mode, the rated power was limited to 143 kW by considerable throttling to reduce the knocking probability. Numerical Results. The characteristics of the examined test cases are reported in Table 4. The simulation of the whole working cycle allows one to account for the real flow field in the engine. Concerning this paper, simulations start from the bottom dead center (BDC) of the exhaust stroke; thus, the whole scavenging phase can be represented together with the effects of the real geometry on the resulting flow field. The strong swirl effect, due to the shape of the intake duct, widely influences all the subsequent processes starting from the evolution of the spray within the chamber and the evaporation process (Figures 1 and 2). The zones where most of diesel vapor (n-heptane) is localized are in fact distorted from the spray direction by the flow field.



Figure 2. Effect of the swirl motion on the evolution of the spray (MF3, CAD 362).

Figure 5. Comparison between experimental and numerical pressure data (MF6, CFM).

Figure 3. Comparison between experimental and numerical pressure data (MF3, CTC).

Figure 6. Comparison between CTC model and CFM (MF6).

Figure 7. Temperature map at 368 CAD (MF6, CTC). Figure 4. Comparison between experimental and numerical pressure data (MF6, CTC).

Most of the fuel vaporizes near the walls of the piston bowl, and ignition first occurs in that zone at the TDC (360 CAD), just 10 CAD after the injection start. Figures 3 and 4 represent the numerical data using the CTC model compared with the experimental ones. The modified CTC model describes well the overall combustion, but slight differences can be observed with respect to the experimental curves. The MF3 case is well represented during the former stage (mainly diesel combustion), whereas the latter one (NG combustion) slightly undervalues the experimental data. The MF6 case is well represented too; nevertheless, a slight difference can be detected during the main combustion. Using a CFM approach (Figure 5), the flame propagation mechanism through the homogeneous mixture can be well described, even better than with the CTC model (Figure 6). Nevertheless, there is an evident undervaluation of the experimental data, mainly as a result of having neglected the contribution of diesel fuel to combustion. Such an error, mainly

localized just after the autoignition of the diesel fuel, widely affects the overall heat release. However, when higher NG utilization (95–98%) is provided, the CTC model could fail and the CFM could better represent combustion. The different behavior of the two models here tested can be better understood through Figures 7-12. By means of the CTC model, the combustion of the whole vaporized diesel fuel and the overall NG can be represented, even though the flame is mainly localized near the ignition points (Figures 79), and the unburned mixture is drawn toward the flame by diffusion. On the contrary, by the CFM, a flame develops from an ignition point and propagates rapidly throughout the chamber (Figures 1012). The flame propagation mechanism highly depends on the air/NG equivalence ratio. An interesting future development could be to define a hybrid model and to switch between CTC and CFM on the basis of the local diesel/NG proportion and the local air/NG equivalence ratio. Moreover, the CFM (which is a typical “flamelet” model) could lose its validity for very lean air/NG


Cordiner et al.

Figure 8. Temperature map at 372 CAD (MF6, CTC).

Figure 11. Temperature map at 372 CAD (MF6, CFM).

Figure 9. Temperature map at 376 CAD (MF6, CTC).



mixtures. Thus, a more general combustion model (G-equation or EDC) could be considered to represent flame propagation. Conclusion Dual-fuel technology has great potentialities to reduce PM and NOx emissions of conventional diesel engines. It is particularly interesting as a retrofit concept because the engine can be easily converted, without great modifications, just by introducing the feeding system for natural gas and modifying the diesel injection pump control. Unfortunately dual-fuel engines suffer from high HC and CO emissions, mainly as a result of the lean combustion of a homogeneous charge. With the introduction of a certain number of devices (intake throttling,

three-way catalyst, EGR) and operating on full diesel at light loads, it is possible to obtain interesting results on the ECE R49 13-mode test compared with the diesel baseline case. Numerical simulation can be a useful tool to help the experimental procedure to find the right direction. In this paper, the first experimental results related to the conversion of a heavy-duty diesel engine to dual-fuel operation are reported, and a numerical procedure to correctly describe the dual-fuel combustion is presented. A mixed 1D-3D numerical procedure was utilized. The 1D code provided the pressure boundary conditions as input information for 3D simulations. Concerning the 3D tools used, a modified version of the KIVA-3V code was utilized, and submodels like WAVE and Shell were introduced to describe the spray evolution and the autoignition process. Moreover, the CTC model, extended to dual-fuel operations, was implemented to simulate the overall combustion. The first results agree well with the experimental data, even though it is not the proper model to represent the flame propagation mechanism. For a high degree of NG utilization (>95%), the CFM can represent a valid alternative. Acknowledgment. The authors wish to thank Marco Pece for his aid in the utilization of numerical tools and the technicians Vincenzo Bonanno and Giuseppe Barese for their assistance with the experimental tests. EF7004755

Numerical and Experimental Analysis of ... - ACS Publications

Recommend Documents