CFD Study and Experimental Validation of a Dual Fuel Engine: Effect of Engine Speed

energies

Article CFD Study and Experimental Validation of a Dual Fuel Engine: Effect of Engine Speed

Roberta De Robbio 1, Maria Cristina Cameretti 2,*, Ezio Mancaruso 1, Raffaele Tuccillo 2 and Bianca Maria Vaglieco 1

1 Istituto di Scienze e Tecnologie per l’Energia e la Mobilità Sostenibili (CNR), 80125 Napoli, Italy; [email protected] (R.D.R.); [email protected] (E.M.); [email protected] (B.M.V.) 2 Department of Industrial Engineering (D.I.I.), Università di Napoli Federico II, 80125 Napoli, Italy; [email protected] * Correspondence: [email protected]

Abstract: Dual fuel engines induce benefits in terms of pollutant emissions of PM and NOx together with carbon dioxide reduction and being powered by natural gas (mainly methane) characterized by a low C/H ratio. Therefore, using natural gas (NG) in diesel engines can be a viable solution to reevaluate this type of engine and to prevent its disappearance from the automotive market, as it is a well-established technology in both energy and transportation fields. It is characterized by high performance and reliability. Nevertheless, further improvements are needed in terms of the optimization of combustion development, a more efficient oxidation, and a more efficient exploitation of gaseous fuel energy. To this aim, in this work, a CFD numerical methodology is described to

simulate the processes that characterize combustion in a light-duty diesel engine in dual fuel mode by analyzing the effects of the changes in engine speed on the interaction between ﬂuid-dynamics and

Citation: De Robbio, R.; Cameretti, chemistry as well as when the diesel/natural gas ratio changes at constant injected diesel amount. M.C.; Mancaruso, E.; Tuccillo, R.; With the aid of experimental data obtained at the engine test bench on an optically accessible research Vaglieco, B.M. CFD Study and engine, models of a 3D code, i.e., KIVA-3V, were validated. The ability to view images of OH Experimental Validation of a Dual distribution inside the cylinder allowed us to better model the complex combustion phenomenon Fuel Engine: Effect of Engine Speed. of two fuels with very different burning characteristics. The numerical results also deﬁned the Energies 2021, 14, 4307. https:// importance of this free radical that characterizes the areas with the greatest combustion activity. doi.org/10.3390/en14144307 Keywords: dual fuel; CFD analysis; combustion; ultraviolet visible spectroscopy Academic Editor: Jamie W.G. Turner

Received: 19 June 2021 Accepted: 15 July 2021 1. Introduction Published: 17 July 2021 The diesel engine has always played a central role in the energy and transportation

Publisher’s Note: MDPI stays neutral sectors. However, in recent years, a number of limitations related to exhaust emissions with regard to jurisdictional claims in have made it the subject of debate. Indeed, diesel engine reliability and low consumption published maps and institutional afﬁl- have led the academic and industrial world to offer alternatives with new fuel blends as iations. well as innovative injection and combustion techniques to comply with current emission regulations. The dual fuel diesel engine still represents a valid alternative to the traditional engine, especially for the reduction of carbon dioxide. Nonetheless, a typical drawback of this new strategy is the employment of a fuel for which the engine setup is not optimized. Therefore, a deep investigation is needed to outline the processes that characterize the Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. conversion of the chemical energy of a fuel source into mechanical work. This article is an open access article In this work, the authors, who have already been involved in the study of dual fuel for distributed under the terms and several years [1–8], present a computational ﬂuid dynamics (CFD) application to investigate conditions of the Creative Commons the phenomenon of combustion of the two fuels used in dual fuel engines (diesel and Attribution (CC BY) license (https:// natural gas/methane). In a dual fuel engine, diesel fuel is injected in minimal quantities to creativecommons.org/licenses/by/ facilitate the ignition of the mixture in the cylinder consisting of air and natural gas that 4.0/). have been premixed in the intake duct.

Energies 2021, 14, 4307. https://doi.org/10.3390/en14144307 https://www.mdpi.com/journal/energies Energies 2021, 14, 4307 2 of 24

The formation of numerous flame nuclei scattered in the combustion chamber leads to a complex ignition phenomenon that is difficult to characterize. A proper experimental activity supported by three-dimensional modelling can make an interpretation of the phenomenon possible. The purpose of this CFD study is in fact to prove the transferability of the experimental results to real diesel engines operating in dual fuel mode and to obtain more detailed information on the phenomena characterizing the dual fuel combustion. Interest in this topic is demonstrated by the presence of many researchers in the scientific literature who have been studying the dual fuel engine to deepen knowledge of the combustion phenomenon and look for suitable injection strategies and methane rates, making use of numerical methodologies and advanced schemes of chemical kinetics. The authors in [9–11] developed a multi-zone model to predict the combustion process with a detailed kinetic scheme that describes the partial oxidation of the gaseous fuel during the compression and the expansion processes. They also addressed the study of the processes that occur during the ignition delay since the introduction of the mixture reduces the presence of oxygen in the cylinder, reducing the temperature and delaying the combustion compared to a traditional diesel engine. The evaluation of emissions whose reduction represents the primary target of this technology is very important [12,13], as emphasized at the beginning and shown in the next sections. In particular, in [12], the authors analyzed the performance and the emissions in steady state condition, improving the combustion and referring to the New European Driving Cycle. The pilot diesel fuel quantity and the pilot injection timing in dual fuel mode, influ- ences highly the evolution of combustion and pollutant formation, as described in [14], where the resulting carbon monoxide (CO) emission levels were higher than those under normal diesel operation modes that were also at high load. On the other hand, the nitrogen oxide (NOx) emissions decrease by 30% due to a major part of the fuel being burned under lean premixed conditions with a lower local temperature. The authors of [3] investigated the injection timing as well, emphasizing the influence on the combustion development as well as on the total hydrocarbons (THC) and the production of NOx fractions and the reduction of carbon dioxide (CO2). In paper [15], Mancaruso et al. showed that dual-fuel combustion reduces the number of soot particles, and the size is dependent on the combustion evolution. The diameter of the particles is higher with respect to conventional diesel operation at low speed, while at high speed, there is a high formation of particles of a smaller size. Another critical operating condition of a dual fuel engine is the part load, as demonstrated by the authors themselves in [2] and by others in [16,17]. An increase in pollutants can be due to a low penetration of the diesel liquid droplets that evaporate late; this problem could be overcome by increasing the quantity of diesel fuel, as demonstrated in [18]. As previously mentioned, CFD simulation is an important tool for the development and optimization of dual fuel engines. Through CFD, numerous tests can be performed to simulate a wide range of engine loads, variation the injection timing, and the amount of natural gas. Efficient calculation models capable of reproducing experimental cases and anticipating new combustion strategies are therefore necessary. An appropriate chemical scheme has to be implemented for an accurate description of the combustion process. In [18], Hockett et al. presented a chemical kinetic mechanism that consisted of 141 species and 709 reactions in order to simulate the combustion of both fuels (n-heptane and natural gas), while a simpler mechanism of 42 species and 57 reactions is shown in [19]. In recent papers [7,8], CFD results are achieved by using a new kinetic scheme by adding 5 new species and 9 reactions for the autoignition of methane with the KIVA-3V solver for a three-dimensional domain to the ‘one step’ model. Since the authors have already investigated different operating conditions by vary- ing the load level, the start of injection, and the amount of methane introduced to the cylinder [2,3,6], this work is mainly focused on the effect of engine speed on dual fuel Energies 2021, 14, 4307 3 of 24

combustion development. Results showed that increasing the rotational speed can exert a relevant inﬂuence on a number of time-dependent processes such as liquid fuel injection and atomization, the mixing of reactant species, and chemical kinetics; therefore, a proper choice in the injection setting can help to achieve ignition in each zone of the combustion chamber. In this way, it is possible to obtain an efﬁcient use of the chemical and energetic contents of the fuel, with a consequent reduction of the emissions of unburned hydrocarbons in the exhaust. Such effects are analyzed via a CFD based approach aiming at the comparison between two different engine regimes. Furthermore, an experimental validation is also attained through the comparison of numerical results with experimental tests conducted on a single-cylinder research engine.

2. CFD Methodology As stated above, the computational activities described in this work were conducted with the purpose of analyzing the effect of rotational speed on the phenomena that characterize dual fuel combustion. This objective is pursued by means of a well-established methodology that was defined in previous work conducted by multiple authors that is briefly recalled in the following paragraphs. The models used for these calculations have already proven to be suitable for the prediction of dual fuel operating conditions [7,8] in terms of pressure, combustion timing, and especially methane ignition. In addition, in this work, the availability of in-cylinder visualizations can represent an important validation tool for the assessment of the reliability of these models and for the description of 3D species distributions. Simulations are performed with the KIVA-3V code [20] on a computational domain consisting of the in-cylinder volume, the external ducts, and the valve lift laws, as illustrated in Figure1. It is important to mention that the geometry of the CFD domain is similar but not identical to the one of the single-cylinder research engine employed for the experiments. The dimensions of the computed engine are reported in Table1. As a matter of fact, the main difference between the two engines (real and simulated) is the bowl shape. The domain used for the numerical calculations is derived from a commercial six-cylinder engine; therefore, this methodology can allow the extension of the experimental results to practical applications. Hence, while the simulated engine is in the shape of a classic Energies 2021, 14, 4307 Mexican hat bowl, the research optical engine has a flat bowl shape due to the presence4 of of25

the sapphire window.

FigureFigure 1.1. ComputationalComputational domaindomain forfor KIVAKIVA code. code.

The kinetic mechanism consists of 18 species (Table 2) and 12 reactions (Table 3). Do- decane (C12H26) is used as a diesel oil surrogate, and its oxidation is represented by a one- step reaction (RR1). The natural gas composition is assumed to be 100% methane, the oxidation of which occurs following two parallel paths represented by low temperature reactions (from RR3 to RR11 in Table 3) and the traction of the RR12 activated below and above the threshold value of 1300 K, respectively. In both cases, oxidation is completed through the reaction of RR2.

Table 2. Detailed species present in the mechanism.

1 C12H26 7 H2 13 HO2 2 O2 8 O 14 CH3 3 N2 9 N 15 H2O2 4 CO2 10 OH 16 CH2O 5 H2O 11 CO 17 CHO 6 H 12 NO 18 CH4

Table 3. Reaction mechanism.

RR1 +37⁄ 2 → 12 + 13 RR2 + 1⁄ 2 ↔ RR3 + → + RR4 + → + RR5 +→+ RR6 + →+ RR7 +→+ RR8 + →+ RR9 + → + +

RR10 + +→ + RR11 →2+ RR12 +3⁄ 2 →+2

Energies 2021, 14, 4307 4 of 24

Table 1. Computation domain dimensions.

Bore 8.4 cm Stroke 8.9 cm Swept Volume 493.22 cm2 Combustion Bowl 26.4 cm2

The presence of the gas exchange phase resulting from the open valve periods requires the calculation of multiple engine cycles to achieve the numerical convergence to the periodic regime and longer computational time. On the other hand, it allows for the correct estimation of (I) the turbulence flow field, (II) the homogenous charge formation, (III) the volumetric efficiency, and (IV) the amount of residual gases in the cylinder. The multi-block structured Cartesian mesh was obtained by using the ICEM meshing tool. The cylinder and bowl volume discretization are composed of nearly 58,000 and 5000 cells, respectively. The kinetic mechanism consists of 18 species (Table2) and 12 reactions (Table3). Dodecane (C12H26) is used as a diesel oil surrogate, and its oxidation is represented by a one-step reaction (RR1). The natural gas composition is assumed to be 100% methane, the oxidation of which occurs following two parallel paths represented by low temperature reactions (from RR3 to RR11 in Table3) and the traction of the RR12 activated below and above the threshold value of 1300 K, respectively. In both cases, oxidation is completed through the reaction of RR2.

Table 2. Detailed species present in the mechanism.

1 C12H26 7 H2 13 HO2 2 O2 8 O 14 CH3 3 N2 9 N 15 H2O2 4 CO2 10 OH 16 CH2O 5 H2O 11 CO 17 CHO 6 H 12 NO 18 CH4

Table 3. Reaction mechanism.

RR1 C12 H26 + 37/2O2 → 12CO2 + 13H2O RR2 CO + 1/2O2 ↔ CO2 RR3 CH4 + O2 → HO2 + CH3 RR4 CH4 + HO2 → H2O2 + CH3 RR5 CH4 + OH → H2O + CH3 RR6 CH3 + O2 → CH2O + OH RR7 CH2O + OH → CHO + H2O RR8 CHO + O2 → CO + HO2 RR9 CHO + M → CO + H + M RR10 H + O2 + M → HO2 + M RR11 H2O2 → 2OH + M RR12 CH4 + 3/2O2 → CO + 2H2O

It is important to highlight that the group of 9 reactions from RR3 to RR11 is a part of the mechanism presented by Li and Williams [21], which consists of 127 reactions and 31 species. It is capable of predicting methane ignition in dual fuel conditions for temperatures between 1000 K and 2000 K, pressures between 1 bar and 150 bar, and equivalence ratios between 0.4 and 3. However, below 1300 K, only the 9 reactions listed can be considered enabled. Moreover, while the temperature and pressure ranges of this mechanism are compatible with those of a high compression ratio diesel engine, the methane–air equivalence ratio can fall outside the validity limits of the model at engine part loads. However, earlier diesel fuel combustion can increase the local equivalence ratio by consuming the oxygen inside the cylinder. Energies 2021, 14, 4307 5 of 24

As already stated, computations aim at providing a more comprehensive interpretation of the tests and completing the experimental information. To this purpose, a kinetic mechanism including important species such as the OH radical is necessary to assess how they can affect the overall oxidation process. In this regard, trends and distributions of the OH radical produced by the experimental activity has been an important validation tool for numerical models. In fact, this species represents a signiﬁcant signal of combustion activity since its formation occurs in the areas characterized by a high level of combustion intensity. In addition, the well-established set of reactions for the prediction of NOx formation introduced by Zel’dovich [22] and six equilibrium reactions (Table4) are elaborated in the CFD code.

Table 4. Equilibrium reactions.

e.1 H2 ↔ 2H e.2 O2 ↔ 2O e.3 N2 ↔ 2N e.4 O2 + H2 ↔ 2OH e.5 O2 + 2H2O ↔ 4OH e.6 O2 + 2CO ↔ 2CO2

The low equivalence ratios of the test cases that locate the mixture outside the flammability limits prevent an efficient flame propagation and then lead to a combustion model with reactions through a turbulent kinetic interaction, such as the Finite Rate–Eddy Dissi- pation (FRED) model, for the description of the combustion process. The Finite Rate–Eddy Dissipation model compares the kinetic with the turbulent reaction rate, and only the smallest one is chosen as the governing factor of combustion. In the KIVA-3V code, the kinetic constant is calculated via the Arrhenius correlation as follows: n k = AT exp(−Ta/T) (1)

where A is the pre-exponential factor, n is temperature exponent, and Ta is the activation temperature. The related values are provided by Li and Williams as shown in Table5.

Table 5. Values of A, n and Ta [21].

3 Reaction A [mol, cm , s] N [-] Ta [K] RR1 8.00 × 10+11 0.0 15,780 RR2 7.9899 × 10+14 0.0 11,613 RR3 3.980 × 10+10 0.0 28,626 RR4 9.04 × 10+09 0.0 12,401 RR5 1.60 × 10+04 1.83 1395 RR6 3.30 × 10+08 0.0 4498 RR7 3.90 × 10+07 0.89 204.5 RR8 3.00 × 10+09 0.0 0.0 RR9 1.86 × 10+09 −1.0 8552 RR10 6.76 × 10+16 −1.4 0.0 RR11 1.20 × 10+14 0.0 22,901 RR12 1.6596 × 10+15 0.0 20,643

Alongside this, the turbulent reaction rate is expressed according to the Magnussen and Hjertager correlation [23]: ε ω = C min x , x (2) t κ f ox

where x f and xox are the mass fractions of fuel and oxygen, κ the turbulent kinetic energy, ε the turbulent dissipation rate, and C is the proportional factor that depends on both the ﬂow Energies 2021, 14, 4307 6 of 24

conditions and the fuel composition. However, due to the premixing of the methane with air, for the reactions from 3 to 12 in Table3, only the kinetic rate is considered. Consequently, the methane ignition time delay, which is based on a chain of kinetic equations, may result in different angular intervals depending on engine rotational speed. On the other hand, the turbulent governed combustion processes (Equation (2)) are strictly dependent on the specific dissipation rate, ε/κ. As such, the different engine regimes could imply changes in the characteristic times of RR1 and RR2 reactions. Clearly, diesel vapour ignition, and, consequently, methane ignition are influenced by the liquid spray distribution and the atomization of the diesel droplets that are responsible for the flame cores. The breakup of the fuel jet was simulated through the WAVE since this model has demonstrated in previous works [3,24] to be reliable for the simulation of this process.

3. Numerical Test Cases Based on the available experimental data discussed in the next sections, numerical tests were performed at two engine speeds, 1500 and 2000 rpm. The experiments were selected to evaluate the influence of methane presence on the combustion characteristics of an automotive diesel engine operating in dual fuel mode. The methane contribution is defined in terms of energy supply ratio of methane to total energy. It corresponds to the premixing level (Premixed Ratio—RP) quantified according to the following relationship:

mp LHVp RP = × 10 (3) mp LHVp + md LHVd

where mp and md are premixed (methane) and directly injected (diesel) fuels mass ﬂow rates, respectively, whereas LHVp and LHVd are the lower heating values of the methane and diesel fuels. Hence, the premixed ratio equal to zero represents normal diesel operation and an RP greater than zero represents dual fuel operation. The engine operating data used in this work for the calculations and the related boundary conditions are listed in Table6. It is worth noting that for each test case, the operating conditions are signiﬁcantly far from the usual load levels of a diesel engine, as demonstrated by the low values of thermal power and indicated mean effective pressure (IMEP).

Table 6. Operating conditions in computational cases.

Engine Speed, rpm 1500 2000 RP [%] 22 22 Thermal Power (fuel LHV) [kW] 6.125 8.16 IMEP [bar] 3.73 3.13 Trapped Air Mass [mg/cycle] 742 730 Start of injection 6 BTDC 10 BTDC Injected diesel oil [mg/cycle] 8.9 8.9 Diesel inj. duration [µs] 450 450 Trapped Methane Mass [mg/cycle] 2.14 2.14 Methane mass fraction [-] 0.00287 0.002923 Air/Fuel Ratio [-] 67.23 66.12 Fuel/Air Equiv. Ratio [-] 0.224 0.228 Methane/Air Equiv. Ratio [-] 0.0497 0.0505 Pressure [bar] 1.27 1.25 Temperature [◦C] 49 53

In previous works [6,8], the case of RP = 22% at 1500 rpm was widely investigated. Results demonstrated that the numerical models were fully capable of reproducing the experimentally detected phenomena and adding important information to the knowledge of methane ignition. Therefore, a second case with the same premixed ratio (22%) but at 2000 rpm was performed in order to investigate the effects of the engine speed on Energies 2021, 14, 4307 7 of 25

Table 6. Operating conditions in computational cases.

Engine Speed, rpm 1500 2000 RP [%] 22 22 Thermal Power (fuel LHV) [kW] 6.125 8.16 IMEP [bar] 3.73 3.13 Trapped Air Mass [mg/cycle] 742 730 Energies 2021, 14, 4307 7 of 24 Start of injection 6 BTDC 10 BTDC Injected diesel oil [mg/cycle] 8.9 8.9 Diesel inj. duration [µs] 450 450 combustion.Trapped Moreover, Methane a thirdMass case [mg/cycle] at 2000 rpm and with 2.14 a reduced premixed 2.14 ratio (18%) was comparedMethane with the mass second fraction one to[-] test the sensitivity of0.00287 the Li and Williams 0.002923 mechanism with different methane/airAir/Fuel Ratio equivalence [-] ratios. For all of 67.23 the investigated strategies, 66.12 the same amount of diesel fuel was injected by means of a single main injection that had the Fuel/Air Equiv. Ratio [-] 0.224 0.228 same duration and injection pressure but a different start in order to compensate for the Methane/Air Equiv. Ratio [-] 0.0497 0.0505 effect of the changes in engine speed. Methane wasPressure injected at[bar] 5 bar in the intake manifold with 1.27 both the intake and 1.25 exhaust valves closed. ThisTemperature means that [°C] the premixed ratio of the intake 49 charge depended 53 solely on the amount of the injected methane. From the values of them measured air and fuel massesMethane during was the injected experiments, at 5 bar it in was the possible intake manifold to obtain with the both mass the fraction intake ofand methane, exhaust sincevalves in closed. the KIVA This code, means this that parameter the premixed is required ratio of to the set intake the inflow charge from depended the intake solely ducts. on Thethe mixtureamount compositionof the injected inside methane. the cylinder From wasthe values calculated of them by the measured code itself air thanks and fuel to themasses presence during of thethe gasexperiments, exchange phase.it was Boundarypossible to conditions obtain the at mass open fraction boundaries, of methane, such as pressuresince in the and KIVA temperature, code, this allowed parameter for is the requ calculationired to set of the mass inflow and from energy the flows intake for ducts. the nextThe computationalmixture composition cycles. inside the cylinder was calculated by the code itself thanks to the presence of the gas exchange phase. Boundary conditions at open boundaries, such as 3.1.pressure Case#1: and RP temperature, = 22% 1500 vs. allowed 2000 rpm for the calculation of mass and energy flows for the nextThe computational two cases thatcycles. are analyzed in this section are characterized by the same relative amount of methane introduced into the intake manifold, which means the same premixed3.1. Case#1: ratio RP (22%)= 22% and 1500 difference vs. 2000 rpm in the engine speed of 1500 and 2000 rpm, respectively. As anticipated,The two cases the that engine are speedanalyzed variation in this inducedsection are changes characterized in the interaction by the same between relative chemistryamount of andmethane in-cylinder introduced fluid-dynamics. into the intake In manifold, particular, which Figure means2 shows the thatsame swirl premixed and tumbleratio (22%) ratios and were difference substantially in the unchanged, engine speed and of this 1500 means and 2000 that therpm, real respectively. swirl and tumble As an- intensitiesticipated, increasedthe engine by speed a 1.33 variation times at 2000induc rpm.ed changes Similarly, in the interaction specific dissipation between rate chemis- that stronglytry and in-cylinder affects the turbulencefluid-dynamics. = chemistry In partic interactionular, Figure appears 2 shows to bethat higher swirl in and the tumble crank angleratiosinterval were substantially close to the unchanged, top dead center and this in the means 2000 that rpm the case real (Figure swirl3 and). Consequently, tumble inten- ansities acceleration increased inby both a 1.33 reactant times mixingat 2000 raterpm. and Similarly, the early the phases specific of dissipation combustion rate can that be expected.strongly affects the turbulence = chemistry interaction appears to be higher in the crank angleIn interval Figure4 close, the comparisonto the top dead of numerical center in the pressures 2000 rpm calculated case (Figure in the 3). two Consequently, cases high- lightsan acceleration that, as expected, in both thereactant increase mixing of the rate engine and speedthe early and phases injection of advancecombustion causes can an be increaseexpected. in the peak pressure.

Figure 2. Comparison of swirl and tumble ratios for two different engine speeds.

Energies 2021, 14, 4307 8 of 25

Figure 2. Comparison of swirl and tumble ratios for two different engine speeds. Energies 2021, 14, 4307 8 of 25

Energies 2021, 14, 4307 8 of 24

Figure 2. Comparison of swirl and tumble ratios for two different engine speeds.

Figure 3. Comparison of specific turbulent dissipation rate for two different engine speeds.

In Figure 4, the comparison of numerical pressures calculated in the two cases highlights that, as expected, the increase of the engine speed and injection advance causes an FigureFigureincrease 3.3. Comparisonin the peak ofpressure. speciﬁcspecific turbulent turbulent dissipationdissipation raterate forfor twotwo differentdifferent engineengine speeds.speeds.

In Figure 4, the comparison of numerical pressures calculated in the two cases highlights that, as expected, the increase of the engine speed and injection advance causes an increase in the peak pressure.

FigureFigure 4.4. ComparisonComparison ofof in-cylinderin-cylinder pressurepressure curvescurves inin RPRP == 2222 @2000@2000 andand 15001500 rpmrpm cases.cases.

Indeed,Indeed, inin orderorder toto ensureensure sufﬁcientsufficient timetime forfor thethe completioncompletion ofof thethe kinetickinetic reactions,reactions, thethe startstart ofof thethe dieseldiesel oiloil injectioninjection waswas advancedadvanced inin thethe casecase withwith higherhigher engineengine speed.speed. InIn addition,Figureaddition, 4. Comparison thethe injectioninjection of startin-cylindestart (SOI)(SOI)r pressure valuesvalues ofofcurves thethe liquid liquidin RP = fuelfuel 22 @2000 reportedreported and 1500 inin TableTable rpm5 cases.5 refer refer to to the the currentcurrent signalsignal provided by the the experimental experimental acti activity.vity. As As a amatter matter of of fact, fact, it itis isimportant important to tomention mentionIndeed, that that ina delayorder a delay ofto 300ensure of 300µs is µsufficient sdetected is detected timebetween betweenfor the the completion current the current signal of signal theand kinetic the and actual the reactions, actual open- openingtheing startof the ofof injector. thethe injector.diesel Hence, oil Hence, injection the actual the actualwas SOI advanc are SOI 3.3 areed BTDC 3.3in the BTDC and case 6.4 and with BTDC. 6.4 higher BTDC. This engine Thismeans means speed. that thatthey In theyaddition,are shifted are shifted the by injection 2.7 by deg 2.7 deg andstart and 3.6 (SOI) 3.6deg values deg at 1500 at 1500of and the and 2000liquid 2000 rpm, fuel rpm, respectively. reported respectively. in Table The The evolution 5 evolution refer to ofthe of a acurrentmultiphase multiphase signal flow ﬂow provided pattern pattern by is is displayedthe displayed experimental in inFigure Figure acti 5;vity.5 in; in both As both acases, matter cases, it is itof ispossible fact, possible it is to important toobserve observe the to thementionspreading spreading that of athe ofdelay theseven sevenof 300liquid liquidµs jets,is detected jets,and andthe be prog thetween progressiveressive the current decrease decrease signal of droplet and of droplet the presence actual presence open- with withingthe ofcorresponding the the corresponding injector. Hence, enlargement enlargement the actual of th SOI ofe regions the are regions 3.3 with BTDC with vaporized and vaporized 6.4 BTDC. fuel. fuel. This means that they are shiftedClearly, by a 2.7 different deg and start 3.6 degof injection at 1500 andimp 2000lies a rpm, different respectively. correspondence The evolution of time of ata multiphaseeach crank angle. flow pattern However, is displayed at 10 crank in angleFigure degree 5; in both after cases, top dead it is possiblecenter (ATDC), to observe almost the spreading of the seven liquid jets, and the progressive decrease of droplet presence with the corresponding enlargement of the regions with vaporized fuel. Clearly, a different start of injection implies a different correspondence of time at each crank angle. However, at 10 crank angle degree after top dead center (ATDC), almost

Energies 2021, 14, 4307 9 of 25

the same amount of time had passed since the injection start (≈1.37 ms at 2000 rpm against ≈1.48 ms at 1500 rpm), and in both cases, an almost complete vaporization of liquid parti- Energies 2021, 14, 4307cles was achieved, and the diesel vapour distributions involved the same areas. This con- 9 of 24 firms that engine speed affects the swirl motion intensity and, consequently, a faster atomization, vaporization, and diesel vapour consumption are obtained.

RP 22% @1500 rpm @2000 rpm

5° BTDC 5° BTDC

TDC TDC

5° 5°

ATDC ATDC

10° 10°

ATDC ATDC

Figure 5. Liquid jet and Figurediesel vapour 5. Liquid distributions. jet and diesel vapour distributions.

Clearly, a different start of injection implies a different correspondence of time at each crank angle. However, at 10 crank angle degree after top dead center (ATDC), almost the same amount of time had passed since the injection start (≈1.37 ms at 2000 rpm against ≈1.48 ms at 1500 rpm), and in both cases, an almost complete vaporization of liquid particles was achieved, and the diesel vapour distributions involved the same areas. This Energies 2021, 14, 4307 10 of 24

Energies 2021, 14, 4307 10 of 25 conﬁrms that engine speed affects the swirl motion intensity and, consequently, a faster atomization, vaporization, and diesel vapour consumption are obtained. Temperature distributionsdistributions in in Figure Figure6 outline 6 outline that, that, as observed as observed for the for pressure the pressure trends, an increase in engine speed leads to a higher peak temperature, although the same regions trends, an increase in engine speed leads to a higher peak temperature, although the same are affected by the higher temperatures. regions are affected by the higher temperatures.

RP 22% @ 1500 rpm @ 2000 rpm

5° ATDC 5° ATDC

10° ATDC 10° ATDC

15° ATDC 15° ATDC

[K]

25° ATDC 25° ATDC

40° ATDC 40° ATDC

Figure 6. Temperature distributions in RP = 22 @1500 and 2000rpm cases. Figure 6. Temperature distributions in RP = 22 @1500 and 2000 rpm cases.

InIn Figure Figure 77a,a, thethe massmass fractionsfractions ofof both fuels are reported for the two examined cases. As previously mentioned, the development of diesel vapour consumption is difficult difﬁcult to evaluate due toto thethe different different engine engine speed speed and and injection injection advance. advance. However, However, in the in 1500the 1500 rpm rpmcase, case, the diesel the diesel vapour vapour starts star to burnts to burn later, later, as demonstrated as demonstrated by the by lower the lower slope ofslope the of curve the curvewithrespect with respect to the 2000to the rpm 2000 case, rpm and case, a slowerand a slower combustion combustion is observed, is observed, as both as curves both curvesoccupy occupy the same the angular same angular interval interval that corresponds that corresponds to a longer to timea longer interval time in interval the 1500 in rpm the 1500case. rpm In both case. cases, In both it is cases, possible it is to possible state that to methanestate that combustionmethane combustion is only activated is only afteracti- vateddiesel after vapour diesel ignition. vapour A zoomignition. of theA zoom methane of the mass methane fraction mass is displayed fraction is in displayed Figure7b inin Figureorder to7b better in order display to better the maindisplay features the main of methanefeatures of combustion. methane combustion. Indeed, as evidencedIndeed, as evidencedby the dashed by the circle, dashed a slight circle, decrease a slight indecrease methane in contentmethane occurs content before occurs the before real ignition, the real ignition,and this and may this be duemay to be the due activation to the activation of the low of the temperature low temperature reactions. reactions. This behaviour This be- haviourwas already was observedalready observed in [7]. In in addition, [7]. In addition, as illustrated as illustrated in Figure in8 Figure, the same 8, the species, same spe- i.e., cies, i.e., CH3, H2O2 and CH2O, involved in the first reactions of the Li and Williams scheme are activated before diesel oil ignition.

Energies 2021, 14, 4307 11 of 24

Energies 2021, 14, 4307 11 of 25 Energies 2021, 14, 4307 11 of 25 CH3,H2O2 and CH2O, involved in the ﬁrst reactions of the Li and Williams scheme are activated before diesel oil ignition.

(a) (b) (a) (b) Figure 7. (a) Mass fractions of n-dodecane and methane comparison for RP = 22% at different engine speeds; (b) methane FigureFigure 7.7. ((aa)) MassMass fractionsfractions ofof n-dodecanen-dodecane andand methanemethane comparisoncomparison forfor RPRP == 22%22% atat differentdifferent engineengine speeds;speeds; ((bb)) methanemethane mass fraction comparison for RP = 22% at different engine speeds. massmass fractionfraction comparisoncomparison forfor RPRP == 22%22% atat differentdifferent engineengine speeds.speeds.

(a) (a)

(b) (b) Figure 8. a b Figure 8. SpeciesSpecies evolution evolution for the RP for the= 22%: RP ( =a) 22%: 1500 ( rpm) 1500 and rpm (b) and2000 ( rpm.) 2000 rpm. Figure 8. Species evolution for the RP = 22%: (a) 1500 rpm and (b) 2000 rpm.

EnergiesEnergies2021 2021, 14, 14, 4307, 4307 12 ofof 2425

Moreover,Moreover, Figure Figure7b 7b displays displays a lowera lower presence presen ofce methaneof methane at the at the end end of combustion of combustion at 2000at 2000 rpm rpm but but also also displays displays a different a different mass mass fraction fraction level level of methane of methane inside inside the the cylinder cylinder as theas the intake intake valve valve closes. closes. This This may may be explained be explained by the by different the different charge charge formation formation estimated esti- duringmated theduring mass the exchange mass exchange phase through phase through the valves. the valves. Additionally,Additionally, it it is is well well known known that that the the volumetric volumetric efficiency efficiency and and the the air air mass mass entering entering thethe cylindercylinder areare affected by by the the engine engine speed. speed. With With the the purpose purpose of identifying of identifying the thecase case that thatpresents presents a more a more efficient efficient consumption consumption of gaseou of gaseouss fuel, fuel, it is itnecessary is necessary to analyze to analyze the frac- the fractiontion of burned of burned methane. methane. AsAs aa matter matter ofof fact, fact, Figure Figure9 9confirms confirms that that the the increase increase of of speed speed can can be be beneficial beneficial for for methanemethane combustion,combustion, asas thethe higher higher burned burned methane methane fraction fraction is is obtained obtained in in the the 2000 2000 rpm rpm case.case. Nevertheless,Nevertheless, thethe samesame figurefigure illustratesillustrates thatthat aa greatgreat amountamount ofof unburnedunburned hydrocar-hydrocar- bonsbons isis present present in in the the exhaust. exhaust. Methane Methane distributions distributions reported reported in in Figure Figure 10 10 demonstrate demonstrate thatthat the the region region where where the the gaseous gaseous fuelfuel isis not not able able to to be be oxidized oxidized is is the the combustion combustion bowl.bowl. InIn this this sense, sense, a a future future optimization optimization of of the the direction direction of of the the liquid liquid fuel fuel jets jets and and the the injection injection timingtiming could could be be appropriate appropriate to to overcome overcome this this issue. issue. InIn Figure Figure 11 11,, the the total total fuel fuel burning burning rate rate (FBR) (FBR) shows shows a a lower lower and and retarded retarded peak peak in in the the 15001500 rpm rpm case, case, as as expected, expected, due due to to the the less less efficient efficient combustion combustion of of methane, methane, evidenced evidenced in in FigureFigure9 ,9, and and due due to to the the lower lower advance advance of of start start of of injection. injection. Finally, Finally, Figure Figure 12 12 displays displays that that higherhigher concentrations concentrations of of nitric nitric monoxide monoxide are are produced produced in in the the 2000 2000 rpm rpm case. case. This This is is mainly mainly duedue toto thethe widerwider zoneszones atat achievedachieved aa higher higher temperature temperaturein in this this context, context,as as previously previously illustratedillustrated in in Figure Figure6 .6.

FigureFigure 9. 9.Methane Methane burnedburned fraction. fraction.

Energies 2021, 14, 4307 13 of 25 Energies 2021, 14, 4307 13 of 24 Energies 2021, 14, 4307 13 of 25

@ 1500 @ 2000 rpm @ 1500 @ 2000 rpm

5° BTDC 5° BTDC 5° BTDC 5° BTDC

TDC TDC TDC TDC

5° ATDC 5° ATDC 5° ATDC 5° ATDC [ppm] [ppm]

10° ATDCATDC 10°10° ATDC ATDC

20° ATDCATDC 20°20° ATDC ATDC

40° ATDC 40° ATDC 40° ATDC 40° ATDC Figure 10. Methane mass fraction contours in RP = 22 @ 1500 and 2000 rpm cases. FigureFigure 10. 10.Methane Methane mass mass fraction fraction contours contours in in RP RP = = 22 22 @ @ 1500 1500 and and 2000 2000 rpm rpm cases. cases.

Figure 11. Total fuel burning rate for the two cases. FigureFigure 11. 11.Total Total fuel fuel burning burning rate rate for for the the two two cases. cases.

Energies 2021, 14, 4307 14 of 24 Energies 2021, 14, 4307 14 of 25

Figure 12. Numerical NO fraction for the two cases. FigureFigure 12.12. NumericalNumerical NONO fractionfraction forfor thethe twotwo cases.cases.

3.2.3.2. Case#2:Case#2: RPRP == 18% vs. 22% at 2000 rpm InIn thisthis sub-section, the the sensitivity sensitivity of of the the Li Liand and Williams Williams mechanism mechanism to different to different me- methanethane equivalence equivalence ratios ratios is istested tested by by comparing comparing the the RP RP = = 22% 22% at 2000 rpm,rpm, whichwhich waswas previouslypreviously analyzed,analyzed, withwith aa casecase operatingoperating atat thethe samesame engineengine speedspeed butbut withwith aa reducedreduced percentagepercentage ofof energyenergy duedue toto methane,methane, suchsuch asas RPRP == 18%.18%. InIn orderorder toto explainexplain thethe results,results, thethe factfact thatthat thatthat thethe differentdifferent premixedpremixed ratiosratios werewere obtainedobtained withwith aa methane methane additionaddition toto thethe baselinebaseline dieseldiesel oiloil amountamount shouldshould bebe considered.considered. ThisThis choicechoice isis substantiallysubstantially differentdifferent fromfrom thethe moremore commoncommon dualdual fuelfuel strategy,strategy, which which consists consists of of a a diesel diesel oil oil substitutionsubstitution withwith methane.methane. TheThe pressurepressure curvecurve forfor RPRP == 18%18% displaysdisplays aa slightslight decreasedecrease withwith respectrespect toto the the RP RP = = 22%22% case,case, asas plottedplotted inin FigureFigure 13 13.. ThisThis waswas expectedexpected sincesince aa reducedreduced amountamount ofof fuel,fuel, i.e.,i.e., methane,methane, waswas introducedintroduced insideinside thethe engine.engine. Nevertheless,Nevertheless, ifif comparedcompared toto thethe 15001500 rpmrpm case,case, whichwhich waswas previouslypreviously illustratedillustrated inin FigureFigure5 ,5, the the pressure pressure level level is is higher higher during during the the entireentire combustioncombustion process. process. This This means means that that the the engine engine speed speed and and the the start start ofof injectioninjection have have a more important effect than the amount of fuel on combustion. a more important effect than the amount of fuel on combustion.

FigureFigure 13.13. ComparisonComparison ofof in-cylinderin-cylinder pressurepressurecurves curvesvarying varyingRP RP and and speed. speed.

Energies 2021, 14, 4307 15 of 24 Energies 2021, 14, 4307 15 of 25

Temperature contours in in Figure Figure 1414 onlyonly show show slight slight differences differences since since combustion combustion is ismainly mainly driven driven by bydiesel diesel oil, oil, and and the the start start of injection of injection and and the theamount amount of liquid of liquid fuel fuel is the is thesame. same. For Forthe thesame same reason, reason, Figure Figure 15a 15 alsoa also does does not not depict depict significant significant differences differences in inn- n-dodecanedodecane consumption. consumption. Additionally, Additionally, in the in the same same figure, figure, it is it possible is possible to notice to notice that that the thecombustion combustion of methane of methane is acti isvated activated by diesel by diesel vapour vapour ignition. ignition. A better A better overview overview of the of themethane methane mass mass fraction, fraction, reported reported in Figure in Figure 15b, shows15b, shows that the that different the different RP values RP values result resultin different in different compositions compositions of the of charge the charge trapped trapped in the in cylinder. the cylinder. As matter As matter of fact, of fact, the thehigher higher concentration concentration of methane of methane in the in theRP = RP 22% = 22% case case provokes provokes higher higher emissions emissions of un- of unburnedburned hydrocarbons, hydrocarbons, despite despite the fact the factthat thatthe RP the = RP18% = case 18% is case characterized is characterized by a slower by a slowerconsumption consumption of methane, of methane, as confirmed as confirmed by Figure by 16. Figure As prev 16.iously As previously mentioned, mentioned, these test thesecases are test characterized cases are characterized by an extremely by an low extremely equivalence low equivalenceratio that locate ratio the that mixture locate out- the mixture outside the flammability limits of methane. However, by comparing the methane side the flammability limits of methane. However, by comparing the methane consump- consumption of RP = 18% with RP = 22% at 1500 rpm, it is possible to observe that tion of RP = 18% with RP = 22% at 1500 rpm, it is possible to observe that increasing the increasing the amount of methane is necessary to promote the development of an efficient amount of methane is necessary to promote the development of an efficient gaseous fuel gaseous fuel combustion. combustion.

RP 22% RP 18%

5° ATDC 5° ATDC

10° ATDC 10° ATDC

15° ATDC 15° ATDC [K]

25° ATDC 25° ATDC

40° ATDC 40° ATDC

Figure 14. Temperature distributions in RP = 22 and 18% at 2000 rpm. Figure 14. Temperature distributions in RP = 22 and 18% at 2000 rpm.

Energies 2021, 14, 4307 16 of 24 EnergiesEnergies 2021 2021, 14, 14, 4307, 4307 1616 of of 25 25

(a(a) ) (b(b) )

FigureFigure 15. 15. (( aa()a) )Mass Mass Mass fractions fractions fractions of of of n-dodecane n-dodecane n-dodecane and and and methane methane methane co comparison comparisonmparison for for RP RP = =22 22 22 and and and 18% 18% 18% at at at 2000 2000 2000 rpm; rpm; ( b(b)) )Methane Methane Methane mass mass mass fractionfractionfraction comparison comparison comparison for for RP RP = =22 22 22 and and and 18% 18% 18% at at at 2000 2000 2000 rpm. rpm. rpm.

Figure 16. Methane burned fraction for RP = 22 and 18% at 2000 rpm. FigureFigure 16. 16.Methane Methane burnedburned fractionfraction for for RP RP = = 2222 andand 18%18% atat 20002000 rpm.rpm.

Moreover,Moreover,Moreover, asas as previouslypreviously previously mentioned, mentioned, mentioned, the the the Li Li andLi and and Williams Williams Williams reaction reaction reaction mechanism mechanism mechanism activates acti- acti- vatesavates weak a a weak methaneweak methane methane oxidation oxidation oxidation before before combustion,before combus combus whichtion,tion, alsowhich which occurs also also withoccurs occurs the with with reduced the the reduced amountreduced amountofamount methane of of methane inmethane the RP in =in the 18% the RP caseRP = = 18% (Figure18% case case 15 (Fig b).(Figure Allure of15b). 15b). the All above All of of the considerationsthe above above considerations considerations are evidenced are are evidencedinevidenced the methane in in the the distributions methane methane distributions distributions at different at crankat diff different angles,erent crank crank which angles, angles, are illustratedwhich which are are inillustrated illustrated Figure 17 in. in FigureFigureThe 17. 17. total fuel burning rates (FBR) in Figure 18 show a slightly lower and delayed peak in theTheThe RP total =total 18% fuel fuel case, burning burning which rates rates can likely(FBR) (FBR) bein in explainedFigure Figure 18 18 byshow show the a reduceda slightly slightly amountlower lower and and of methanedelayed delayed peakintroducedpeak in in the the RP in RP the= = 18% 18% cylinder. case, case, which Forwhich the can can same likely likely reason, be be explained explained the mass by fractionsby the the reduced reduced of nitric amount amount monoxide of of me- me- in thaneFigurethane introduced introduced19 show that in in the thethe cylinder. reductioncylinder. For For of the methanethe same same reason, inreason, the RP the the = mass mass 18% fractions case fractions causes of of nitric anitric decrease monox- monox- of idethiside in pollutantin Figure Figure 19 19 species show show that in that the the the exhaust. reduction reduction of of me methanethane in in the the RP RP = = 18% 18% case case causes causes a a decrease decrease ofof this this pollutant pollutant species species in in the the exhaust. exhaust.

Energies 2021, 14, 4307 17 of 25 Energies 2021, 14, 4307 17 of 24 Energies 2021, 14, 4307 17 of 25

RP 22% RP 18% RP 22% RP 18%

TDC TDC TDC TDC

5° ATDC 5° ATDC 5° ATDC 5° ATDC

[ppm][ppm] 10°10° ATDC ATDC 10°10° ATDC ATDC

20°20° ATDC ATDC 20°20° ATDC ATDC

40°40° ATDC ATDC 40°40° ATDC ATDC Figure 17. Methane mass fraction contours for RP = 22 and 18% at 2000 rpm. FigureFigure 17. 17.Methane Methane mass mass fraction fraction contours contours for for RP RP = = 22 22 and and 18% 18% at at 2000 2000 rpm. rpm.

Figure 18. Total fuel burning rate for the RP = 22 and 18% at 2000 rpm cases. FigureFigure 18. 18.Total Total fuel fuel burning burning rate rate for for the the RP RP = = 22 22 and and 18% 18% at at 2000 2000 rpm rpm cases. cases.

Energies 2021, 14, 4307 18 of 24 Energies 2021, 14, 4307 18 of 25 Energies 2021, 14, 4307 18 of 25

FigureFigureFigure 19.19. 19. Numerical Numerical NO NO fraction fraction for for RP RP = = 22= 22 22 and and and 18% 18% 18% at at at2000 2000 2000 rpm rpm rpm cases. cases. cases.

Finally,Finally,Finally, thethe the OHOH OH radical radical is is an an important important species species species in in in the the the Li Li Li and and and Williams Williams Williams kinetic kinetic kinetic mecha- mecha- mecha- nismnismnism (Table(Table (Table3 ),3), 3), but but but this this this new new new scheme scheme scheme only only only describes de describesscribes the the the methane methane methane ignition ignition ignition process. process. process. Moreover, Moreo- Moreo- thever,ver, oxidationthe the oxidation oxidation of the of of the gaseousthe gaseous gaseous fuel fuel fuel is is activated is acti activatedvated by by by the the the firstfirst first threethree three reactions reactions (RR3 (RR3 (RR3 to to toRR5) RR5) RR5) thatthatthat dodo do notnot not presentpresent present thisthis this radical radical among among the the the pr products; products;oducts; only only only reaction reaction reaction RR5 RR5 RR5 involves involves involves OH OH OH as as asa a a reactantreactantreactant resultingresulting resulting from from the the equilibrium equilibrium equati equations equationsons e.4 e.4 e.4 and and and e.5 e.5 e.5 from from from Table Table Table 4.4. 4.The The The contours contours contours ofofof the the OH OHOH mass mass fractions fractions fractions in in inFigure Figure Figure 20 20 illustrate20 illustrate illustrate that that thatthis this radical this radical radical is islocated located is located in inzones zones in at zones higherat higher at highertemperaturestemperatures temperatures previously previously previously already already already discussed discussed discussed in in Figure Figure in Figure 14, 14, confirming confirming 14, confirming that that this thatthis species thisspecies species may may mayrepresentrepresent represent an an important important an important marker marker marker of of the the ofintensity intensity the intensity of of combustion. combustion. of combustion. As As also also Assuggested suggested also suggested by by Figure Figure by Figure21,21, higher higher 21, higherconcentrations concentrations concentrations are are observed observed are observed in in th the eRP inRP = the =22% 22% RP case =case 22% because because case OH because OH is isonly only OH present ispresent only presentinin the the kinetic kinetic in the kineticmechanism mechanism mechanism of of methane, methane, of methane, and and it itis and isexpected expected it is expected to to be be moreto more be present more present present in incases cases in with cases with withaa higher higher a higher methane methane methane supply. supply. supply. Moreover, Moreover, Moreover, it itis is inte itinte isresting interestingresting to to notice notice to notice how how the how the distributions thedistributions distributions pre- pre- presentsentsent the the thesame same same shape: shape: shape: the the two the two twocases cases cases being being being perf performed performedormed at at the the at same the same same engine engine engine speed speed speed confirm confirm confirm that that thatthethe same thesame same swirl swirl swirl motion motion motion has has hasbeen been been established established established in insideside inside the the cylinder, the cylinder, cylinder, show show show the the theinvolvement involvement involvement of of ofthethe the same same same regions regions regions in in inthe the the oxidation oxidation oxidation process. process. process.

RPRP 22% 22% RP RP 18% 18%

5°5° ATDC ATDC 5°5° ATDC ATDC

10°10° ATDC ATDC 10°10° ATDC ATDC

[-][-]

20°20° ATDC ATDC 20°20° ATDC ATDC Figure 20. OH mass fraction contours for RP = 22 and 18% at 2000 rpm. FigureFigure 20. 20.OH OH mass mass fraction fraction contours contours for for RP RP = = 22 22 and and 18% 18% at at 2000 2000 rpm. rpm.

Energies 2021, 14, 4307 19 of 24 Energies 2021, 14, 4307 19 of 25

RP 22% RP 18%

5° ATDC 5° ATDC

10° ATDC 10° ATDC

[-]

20° ATDC 20° ATDC

Figure 21. OH mass fraction contours for RP = 22 and 18% at 2000 rpm in a transversal plane. Figure 21. OH mass fraction contours for RP = 22 and 18% at 2000 rpm in a transversal plane.

4.4. Experimental Experimental ActivityActivity TheThe experimentalexperimental activity activity was was conducted conducted on ona single a single cylinder, cylinder, four-stroke, four-stroke, direct direct in- injection,jection, diesel diesel engine, engine, with with a amulti multi valve valve prod productionuction head head of ofa 2 a L 2 Lengine. engine. The The engine engine is is ableable to to run run continuously.continuously. Table7 7 reports reports thethe specificationsspecifications of of the the engine engine and and the the injection injection system.system. The production head was modified for the single cylinder research engine with particular attention to the water-cooling and lubricating oil head ducts. The transparent en- Table 7. Engine and injection systems specifications. gine utilized a classic extended piston with piston crown window with a diameter of 46 mm, which wasEngine made Typeof sapphire. Using a 45° UV–VISible 4-Stroke mirror Single inside Cylinder the extended piston, a full view of the combustion bowl was detected (Figure 22). Bore 8.5 cm Stroke 9.2 cm Table 7. EngineConnecting and injection rod lengthsystems specifications. 16 cm 3 EngineSwept volume Type 4-Stroke Single522 cm Cylinder Combustion bowl 19.7 cm3 Bore 8.5 cm Vol. compression ratio 16.5:1 Stroke 9.2 cm Diesel injection system Common rail DI Connecting rod length 16 cm Injector type Solenoid driven, Minisac nozzle 3 SweptNumber volume of holes 522 cm 7 ConeCombustion angle of fuel bowl jet axis 19.7 148cm3◦ Vol. compressionHole diameter ratio 16.5:1 0.141 mm Rated flow @ 100 bar 440 cm3/30 s

Energies 2021, 14, 4307 20 of 24

Energies 2021, 14, 4307 20 of 25

Table 7. Cont.

Engine Type 4-Stroke Single Cylinder Diesel injection system Common rail DI Methane injection system PFI NumberInjector of type holes Solenoid driven, 1 Minisac nozzle MaximumNumber injection of holes pressure 5 7bar ConeRated angle ﬂow of fuel @ 3 barjet axis 1600 148° cm3/min Hole diameter 0.141 mm The productionRated flow head @ 100 was bar modiﬁed for the single cylinder 440 research cm3/30 engine s with particular attentionMethane to the injection water-cooling system and lubricating oil head ducts.PFI The transparent engine utilized a classicNumber extended of holes piston with piston crown window with 1 a diameter of 46 mm, ◦ which wasMaximum made of injection sapphire. pressure Using a 45 UV–VISible mirror inside 5 bar the extended piston, a full view of theRated combustion flow @ 3 bar bowl was detected (Figure 22). 1600 cm3/min

FigureFigure 22.22. ExperimentalExperimental apparatus.apparatus.

TheThe engineengine waswas equippedequipped withwith commoncommon railrail injectioninjection systemsystemcapable capableof ofa amaximum maximum injectioninjection pressure pressure of of 1800 1800 bar. bar. The The injector injector with with a solenoid a solenoid valve valve was was centrally centrally located, located, had thehad same the same cylinder cylinder axis, andaxis, mounted and mounted a single a single guide guide microsac microsac nozzle. nozzle. The nozzle The nozzle had seven had holesseven that holes were that symmetrically were symmetrically distributed, distribu and eachted, and hole each had ahole convergent-divergent had a convergent-diver- shape. Thegent number shape. ofThe injections number (upof injections to 5) per cycle,(up to the 5) startper cycle, of injection the start (SOI), of injection and the energizing (SOI), and timingthe energizing of injection timing (ET) of as injection well as the (ET) dwell as we timell as between the dwell the time consecutive between injectionsthe consecutive were managedinjectionsby were a fully managed flexible by electronic a fully flexible control electronic unit (ECU). control In Table unit7 ,(ECU). the injection In Table system 7, the specificationsinjection system are reported.specifications On the are other reported. hand, theOn production the other intakehand, manifoldthe production of the engine intake wasmanifold modified of the to engine add and was set modified an electronic to add injector and set that an electronic is commonly injector used that in modern is commonly port fuelused injection in modern (PFI) port systems fuel injection of methane (PFI) engines. systems It wasof methane applied engine directlys. intoIt was the applied non-swirl di- actuatedrectly into intake the non-swirl port, where actuated it injected intake pure port, methane where fuel.it injected The PFI pure system methane was fedfuel. by The a compressedPFI system methanewas fed bottleby a compressed at a working methane pressure bottle of 5 bar. at Aa working delay unit pressure synchronized of 5 bar. with A thedelay engine unit shaftsynchronized encoder was with used the engine to control shaft methane encoder injection was used in termsto control of SOI methane and ET. injec- The delaytion in unit terms generated of SOI and a pulse ET. thatThe manageddelay unit the generated PFI injection a pulse timing. that Themanaged desired the amount PFI injec- of methanetion timing. fuel The was desired managed amou throughnt of methane the pulse fuel width was for managed each injection through event. the pulse width for eachIn the injection experimental event. apparatus, sensors were installed to control the following parame- ters overIn the time: experimental the engine apparatus, temperature, sensors the intakewere installed air temperature to control and the pressure, following and param- the exhausteters over gas time: temperature the engine and temperature, pressure. The the Sensyflowintake air temperature FMT500-IG wasand usedpressure, to measure and the theexhaust air mass gas flowtemperature rate. More and details pressure. and specificationsThe Sensyflow of FMT500-IG the engine are was reported used to in measure ref. [5]. the airFor mass each operatingflow rate. conditionMore details that and was investigated,specifications the of cylinderthe engine pressure are reported and the in drive ref. injector[5]. current signals were digitalized and recorded every 0.2 crank angle (approximately) degrees.For Theeach average operating in-cylinder condition pressure that was was investigated, calculated from the 200 cylinder consecutive pressure combustion and the cycles.drive injector The heat current release signals rate was were calculated digitalized from and the recorded averaged every pressure 0.2 crank data viaangle the (approx- typical first law and perfect gas analysis [25]. During the test, the engine was stopped, and the imately) degrees. The average in-cylinder pressure was calculated from 200 consecutive combustion cycles. The heat release rate was calculated from the averaged pressure data via the typical first law and perfect gas analysis [25]. During the test, the engine was

Energies 2021, 14, 4307 21 of 24

piston window was cleaned. Table8 shows the properties of the diesel and methane fuels used in the tests.

Table 8. Chemical and physical properties for diesel and methane fuels.

Feature Diesel Methane

Chemical formula CnH1.8n CH4 Density @ 15 ◦C [kg/m3] 833.1 0.678 Viscosity at 40 ◦C [mm2/s] 3.141 - Cetane number 51.8 - Research Octane Number - >120 Lower heating value [MJ/kg] 42.97 50 Auto-ignition temperature [◦C] 220 650 Flammability limits [% vol] 0.6–7.5 5–15 A/F st 14.54 17.24 Carbon [%, mol/mol] 85.22 75 Hydrogen [%, mol/mol] 13.03 25 Nitrogen [%, mol/mol] 0.04 - Oxygen [%, mol/mol] 1.45 -

Figure 22 displays the experimental set-up used for the digital imaging analysis to perform the natural emission measurements of the ﬂame intensity in the ultraviolet (UV) and visible (VIS) range, respectively. Spectroscopic measurements were conducted by means of a gated intensiﬁed CCD (ICCD) camera and a spectrograph at high sensitivity in the UV–VIS range. Table9 shows the prominent spectral peaks of the intermediates and products of combustion in the UV and VIS regions [26].

Table 9. Emission peaks of major combustion species determined in the engine.

CH (nm) OH (nm) CH2O (nm) C2 (nm) HCO (nm) CN (nm) 314 368 470–474 320, 330 358 387–389 302–309 * 384 516 * 330, 340 387, 388 431 * 395 558–563 355, 360 412–457 380, 385 An asterisk symbol highlights the strongest emission bands.

The spectroscopic measurements detected the temporal evolutions and the spatial distribution of the UV–VIS emission light of the typical combustion species. The images were recorded with an ICCD camera. In particular, the OH spatial distribution was investigated by applying a band pass ﬁlter at the 310 nm wavelength in front of the objective lens [26]. The ﬁlter had a band width of +/− 5 nm. Only one image per engine combustion cycle was acquired with a step of approximately 1◦. The ICCD camera synchronization with the engine was obtained by means of a delay unit. In particular, the images and spectra were detected with an exposure time of 55 and 41 µs. They correspond to a 0.5◦crank angle at engine speeds of 1500 and 2000 rpm, respectively.

5. Experimental Validation In this last section, a comparison of the numerical results with the experimental measurements is necessary to evaluate the reliability of the Li and Williams model to describe methane ignition in different operating conditions. The calculated in-cylinder pressures reported in Figure 23 show a fair compliance with the experimental measurements in both the RP = 22% and 18% cases, especially regarding the compression and expansion phases. Besides the divergence near the TDC, the rise to the peak and the peak are also correctly detected both in terms of the crank angle and the pressure level. However, it is important to highlight an increase in pressure near the top dead center before the real rise to the peak was detected in the experimental activity. This detail could be due to the low temperature reactions already evidenced in the diagrams Energies 2021, 14, 4307 22 of 25

and the pressure level. However, it is important to highlight an increase in pressure near the top dead center before the real rise to the peak was detected in the experimental activity. This detail could be due to the low temperature reactions already evidenced in the diagrams of the methane mass fractions (Figures 7b and 15b). This means that the Li and Williams model seems capable of simulating the correct methane ignition, although the intensity of these reactions cannot be observed in the pressure curve, and perhaps this is not sufficient to provoke the pressure increase near the TDC. Figure 24 compares the calculated total fuel burning rate and the OH measured during the experimental activity, both non-dimensional, since they are two different varia- Energies 2021, 14, 4307 22 of 25 bles. Nevertheless, it should be noted that the start of FBR perfectly matches the start of Energies 2021, 14, 4307 22 of 24 OH radical formation. It is likely that the low temperature reactions do not include the formation of this radical, as is already highlighted in Figure 8. In addition, the peaks of experimentaland the pressure OH level.are anticipated However, atit isonly importan about t1 to crank highlight angle an degree increase with in respectpressure to near the peaksofthe the top methaneof dead fuel centerburning mass before fractions rates. the Since real (Figures the rise Total to7b the and FBR peak 15 takesb). was This the detected means consumption in that the the experimental of Li both and Williamsfuels activ- into account,modelity. This seems thisdetail may capable could be due ofbe simulating dueto the to fullthe the activationlow correct temperature methaneof methane reactions ignition, combustion. already although Finally, evidenced the intensity the dura-in the of tionthesediagrams of reactions both of combustion the cannot methane be and observedmass OH fractions radical in the pres(Figures pressureence 7bshows curve, and anda15b). strong perhaps This relationship. means this is that not Therefore,the sufﬁcient Li and basedtoWilliams provoke on thesemodel the pressureassumptions, seems capable increase it is of nearpossible simulating the TDC.to affirm the correct that the methane hydroxyl ignition, radical althoughOH can rep- the resentintensity an importantof these reactions signal for cannot the description be observed and in developmentthe pressure curve, of dual and fuel perhaps combustion. this is not sufficient to provoke the pressure increase near the TDC. Figure 24 compares the calculated total fuel burning rate and the OH measured during the experimental activity, both non-dimensional, since they are two different variables. Nevertheless, it should be noted that the start of FBR perfectly matches the start of OH radical formation. It is likely that the low temperature reactions do not include the formation of this radical, as is already highlighted in Figure 8. In addition, the peaks of experimental OH are anticipated at only about 1 crank angle degree with respect to the peaks of fuel burning rates. Since the Total FBR takes the consumption of both fuels into account, this may be due to the full activation of methane combustion. Finally, the duration of both combustion and OH radical presence shows a strong relationship. Therefore, based on these assumptions, it is possible to affirm that the hydroxyl radical OH can represent an important signal for the description and development of dual fuel combustion.

(a) (b)

FigureFigure 23.23. In-cylinderIn-cylinder pressurepressure curvecurve inin ((aa)) RPRP == 22%22% andand ((bb)) RPRP == 18% cases @ 2000 rpm.

Figure 24 compares the calculated total fuel burning rate and the OH measured during the experimental activity, both non-dimensional, since they are two different variables. Nevertheless, it should be noted that the start of FBR perfectly matches the start of OH radical formation. It is likely that the low temperature reactions do not include the formation of this radical, as is already highlighted in Figure8. In addition, the peaks of experimental OH are anticipated at only about 1 crank angle degree with respect to the peaks of fuel burning rates. Since the Total FBR takes the consumption of both fuels into account, this may be due to the full activation of methane combustion. Finally, the duration of both combustion and OH radical presence shows a strong relationship. Therefore, based on these(a) assumptions, it is possible to afﬁrm that the hydroxyl(b) radical OH can represent an Figure 23. In-cylinderimportant pressure signal for curve the in description (a) RP = 22% and and development (b) RP = 18% cases of dual @ 2000 fuel rpm. combustion.

(a) (b)

Figure 24. Non-dimensional experimental OH and numerical FBR curves in (a) RP = 22% and (b) RP = 18% cases @ 2000 rpm.

Energies 2021, 14, 4307 23 of 24

6. Conclusions Supplying diesel engines with natural gas in dual fuel mode allows the revaluation of this well-established technology by exploiting the advantages of a gaseous fuel to meet environmental requirements. However, a correct understanding and description of the combustion process are necessary for the research of optimization strategies in energy conversion that avoid emissions of unburned hydrocarbons caused by the incomplete oxidation of natural gas and the consequent improper exploitation of its energy content. For these reasons, the effects of the engine speed on dual fuel combustion development are analyzed via a CFD based approach on a light duty dual fuel diesel engine supplied with an increasing amount of methane. The injection of diesel oil is unchanged, except for the start, which has to be anticipated when increasing the speed regime. In particular, two engine speeds (1500 and 2000 rpm) and two different methane energy supplies (RP = 22 and 18%) were investigated. Moreover, the sensitivity of the Li and Williams model for the description of methane ignition has been tested when the engine speed varies the conditions of premixing formation inside the cylinder. Results showed that the swirl and tumble intensities and the specific dissipation rate increased proportionally to the engine speed. These variables strongly affect the turbulence–chemistry interaction by improving the atomization, mixing of reactant species, and chemical kinetics, leading to higher pressure levels and temperature peaks, even when a lower amount of methane was introduced in the cylinder. Indeed, in the 1500 rpm case, the diesel vapour started to burn later, and a slower combustion was observed with respect to the 2000 rpm case. Additionally, the speed increase can be beneficial for methane combustion, as a higher burned methane fraction was obtained in the 2000 rpm case. In addition, a lower presence of methane at the end of combustion at 2000 rpm was observed, as was a different mass fraction level of methane inside the cylinder as the intake valve closed, demonstrating that a different charge formation was estimated during the mass exchange phase through the valves. The Li and Williams model proved its capability to describe the first phase of methane oxidation by detecting a slight decrease before real ignition due to the activation of low temperature reactions at 2000 rpm. Finally, a validation of the simulations was achieved through a comparison with the experimental measurements obtained from an experimental activity conducted on an optically accessible single-cylinder research engine. In particular, the spatial distribution and temporal evolutions of the UV–VIS emission allowed the visualization of the OH radical, which represents an important marker of the intensity and the development of combustion, which was confirmed by the numerical outcomes in terms of the total fuel burning rate.

Author Contributions: Conceptualization, methodology, software, R.T., M.C.C. and R.D.R.; validation, E.M. and B.M.V. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Conﬂicts of Interest: The authors declare no conﬂict of interest.

Nomenclature

ATDC After Top Dead Center BTDC Before Top Dead Center CFD Computational Fluid Dynamics DI Direct Injection ECU Electronic Control Unit FBR Fuel Burning Rate FRED Finite Rate—Eddy Dissipation IMEP Indicated Mean Effective Pressure LHV Lower Heating Value NG Natural Gas Energies 2021, 14, 4307 24 of 24

PFI Port Fuel Injection PM Particulate Matter RP Premixed Ratio SOI Start of Injection THC Total Hydrocarbons

References 1. Abagnale, C.; Cameretti, M.; De Simio, L.; Gambino, M.; Iannaccone, S.; Tuccillo, R. Numerical simulation and experimental test of dual fuel operated diesel engines. Appl. Therm. Eng. 2014, 65, 403–417. [CrossRef] 2. Abagnale, C.; Cameretti, M.C.; Ciaravola, U.; Tuccillo, R.; Iannaccone, S. Dual Fuel Diesel Engine at Variable Operating Conditions: A Numerical and Experimental Study; SAE Technical Paper 2015-24-2411; SAE: Warrendale, PA, USA, 2015. [CrossRef] 3. Cameretti, M.; Tuccillo, R.; De Simio, L.; Iannaccone, S.; Ciaravola, U. A numerical and experimental study of dual fuel diesel engine for different injection timings. Appl. Therm. Eng. 2016, 101, 630–638. [CrossRef] 4. Cameretti, M.C.; De Robbio, R.; Tuccillo, R. Performance Improvement and Emission Control of a Dual Fuel Operated Diesel Engine; SAE Technical Paper 2017-24-0066; SAE: Warrendale, PA, USA, 2017. [CrossRef] 5. Magno, A.; Mancaruso, E.; Vaglieco, B.M. Combustion Analysis of Dual Fuel Operation in Single Cylinder Research Engine Fuelled with Methane and Diesel; SAE Technical Paper 2015-24-2461; SAE: Warrendale, PA, USA, 2015. [CrossRef] 6. Mancaruso, E.; Sequino, L.; Vaglieco, B.M.; Cameretti, M.C.; De Robbio, R.; Tuccillo, R. CFD Analysis of the Combustion Process in Dual-Fuel Diesel Engine; SAE Technical Paper 2018-01-0257; SAE: Warrendale, PA, USA, 2018. [CrossRef] 7. Cameretti, M.C.; De Robbio, R.; Tuccillo, R.; Pedrozo, V.; Zhao, H. Integrated CFD-Experimental Methodology for the Study of a Dual Fuel Heavy Duty Diesel Engine; SAE Technical Paper 2019-24-0093; SAE: Warrendale, PA, USA, 2019. [CrossRef] 8. De Robbio, R.; Cameretti, M.; Tuccillo, R. Ignition and combustion modelling in a dual fuel diesel engine. Propuls. Power Res. 2020, 9, 116–131. [CrossRef] 9. Liu, Z.; Karim, G.A. A Predictive Model for the Combustion Process in Dual Fuel Engines; SAE Technical Paper 952435; SAE: Warrendale, PA, USA, 1995. [CrossRef] 10. Liu, Z.; Karim, G.A. An examination of the ignition delay period in gas-fueled diesel engines. J. Eng. Gas Turbines Power 1998, 120, 225–231. [CrossRef] 11. Karim, G.A.; Liu, Z.; Jones, W. Exhaust Emissions from Dual Fuel Engines at Light Load; SAE Technical Paper 932822; SAE: Warrendale, PA, USA, 1993. [CrossRef] 12. Taniguchi, S.; Masubuchi, M.; Kitano, K.; Mogi, K. Feasibility Study of Exhaust Emissions in a Natural Gas Diesel Dual Fuel (DDF) Engine; SAE Technical Paper 2012-01-1649; SAE: Warrendale, PA, USA, 2012. [CrossRef] 13. Lounici, M.S.; Loubar, K.; Tarabet, L.; Balistrou, M.; Niculescu, D.-C.; Tazerout, M. Towards improvement of natural gas-diesel dual fuel mode: An experimental investigation on performance and exhaust emissions. Energy 2014, 64, 200–211. [CrossRef] 14. Liu, J.; Yang, F.; Wang, H.; Ouyang, M.; Hao, S. Effects of pilot fuel quantity on the emissions characteristics of a CNG/diesel dual fuel engine with optimized pilot injection timing. Appl. Energy 2013, 110, 201–206. [CrossRef] 15. Di Iorio, S.; Magno, A.; Mancaruso, E.; Vaglieco, B.M. Characterization of particle number and mass size distributions from a small compression ignition engine operating in diesel/methane dual fuel mode. Fuel 2016, 180, 613–623. [CrossRef] 16. Mousavi, S.M.; Saray, R.K.; Poorghasemi, K.; Maghbouli, A. A numerical investigation on combustion and emission characteristics of a dual fuel engine at part load condition. Fuel 2016, 166, 309–319. [CrossRef] 17. Papagiannakis, R.G.; Hountalas, D.T.; Krishnan, S.R.; Srinivasan, K.K.; Rakopoulos, D.C.; Rakopoulos, C. Numerical evaluation of the effects of compression ratio and diesel fuel injection timing on the performance and emissions of a fumigated natural gas–diesel dual-fuel engine. J. Energy Eng. 2016, 142, E4015015. [CrossRef] 18. Hockett, A.; Hampson, G.; Marchese, A.J. Development and validation of a reduced chemical kinetic mechanism for computational ﬂuid dynamics simulations of natural gas/diesel dual-fuel engines. Energy Fuels 2015, 30, 2414–2427. [CrossRef] 19. Maghbouli, A.; Saray, R.K.; Shafee, S.; Ghafouri, J. Numerical study of combustion and emission characteristics of dual-fuel engines using 3D-CFD models coupled with chemical kinetics. Fuel 2013, 106, 98–105. [CrossRef] 20. Amsden, A.A. KIVA-III v: Block structured KIVA program Engine with vertical or canted valves, LA—Los Angeles 13313—MS, Los Alamos. 1997. Available online: https://www.osti.gov/biblio/505339-kiva-block-structured-kiva-program-engines-vertical- canted-valves (accessed on 9 July 2021). 21. Li, S.C.; Williams, F.A. Reaction mechanisms for methane ignition. J. Eng. Gas Turbines Power 2002, 124, 471–480. [CrossRef] 22. Zel’dovich, Y.B.; Sadovnikov, P.Y.; Frank-Kamenetskik, D.A. Oxidation of Nitrogen in Combustion; Academy of Science of SR, Institute of Chemical Physics: Moscow-Leningrad, Russia, 1947. 23. Magnussen, B.; Hjertager, B. On mathematical modeling of turbulent combustion with special emphasis on soot formation and combustion. Symp. (Int.) Combust. 1977, 16, 719–729. [CrossRef] 24. Cameretti, M.C.; Tuccillo, R. Flow and atomization models for C.R. diesel CFD simulation. In Proceedings of the ASME/IEEE 2007 Joint Rail Conference and Internal Combustion Engine Division Spring Technical Conference, Pueblo, CO, USA, 13–16 March 2007. Paper No. JRCICE2007-40068. [CrossRef] 25. Heywood, J.B. Internal Combustion Engine Fundamentals; Mc Graw-Hill: New York, NY, USA, 1988. 26. Gaydon, A.G. The Spectroscopy of Flames; Chapman and Hall: London, UK, 1957.