3.1. Thermodynamic Analysis
In-cylinder pressure was analyzed as the average value of 200 consecutive acquisitions for five different fuels and three air-fuel dilutions. The results for stoichiometric and the leanest air-fuel ratio are reported in
Figure 5 and
Figure 6, respectively. The motored pressure signal before the combustion process is depicted with dash-double-dot lines. Pure syngas shows the highest peak pressure in both H
2/CO and air diluted conditions. Despite the inert gas content in real syngas mixtures, the peak pressure was higher than that of methane for all conditions. On the other hand, the pressure trace in the expansion stroke for syngas mixtures (pure and diluted) was lower compared to the baseline fuel. This behavior could be attributed to the spark timing optimization strategy, which has a strong effect on this parameter [
23]. The high difference in combustion duration caused a shift of the peak pressure position to the left. In stoichiometric conditions (
Figure 5) the real syngas featured a shift of around 8 CAD with respect to pure syngas (for the same H
2 proportion). The influence of fuel dilution was similar for both hydrogen levels at λ = 1.0, with a reduction of around 20% in peak pressure. Air dilution exerts a strong effect on combustion behavior, with an important pressure decrease for methane and real syngas mixtures (
Figure 6). The high hydrogen content in pure syngas mixtures enhanced flame propagation and resulted in a reduction of only 5 bar in the peak pressure with respect to stoichiometric conditions.
The mass fraction burned was calculated with the thermodynamic first-law procedure explained in
Section 2.3. Global results of the MFB traces are detailed in
Table 4, with 5%, 10%, and 50% points being representative of the initial, kernel development, and main combustion phases, respectively. The effect of hydrogen on the combustion speed is evident, greatly reducing the duration of each stage. In addition, air dilution also plays an important role, especially for fuels with lower laminar speed (see
Table 3). This information is useful to have the first reference when re-calibration of the ECU is made for this type of alternative fuels.
Other parameters such as IMEP, combustion stability, and fuel conversion efficiency were obtained from the analysis of pressure traces. IMEP is commonly used as a normalized parameter for power analysis.
Figure 7 shows the IMEP for the five fuels and three air-fuel ratios tested. Methane featured the highest value for stoichiometric operation and intermediate lean conditions. However, for the leanest case (λ = 1.4) pure syngas with the H
2/CO ratio of 50/50% IMEP was higher than the baseline. Methane showed an important increase in combustion duration and reduction of peak in-cylinder pressure. This could be explained due to the SA optimization approach, which has a strong effect in lean conditions. On the other hand, the fixed spark advance seems to be better suited for pure syngas mixtures. Practically, the stoichiometric and the lean mixture feature the same IMEP, and thus the same power output.
An important decrease in IMEP was recorded for diluted syngas only at λ = 1.4, mainly due to a combined effect of fuel and air dilution. At this point the losses in volumetric efficiency are high. When comparing the H2 content, it can be seen that the 50% H2 blend has higher power output than the 75% H2 one under the tested conditions. This behavior could be associated with the effect of advanced spark timing and lower volumetric efficiency due to the reduced density of hydrogen. In spite of the fact that the main objective of the test did not include measuring the performance of the fuels, the results suggest that re-calibration of the ECU needs to be performed when considering such a fuel type change and other air management and injection strategy need to be applied. Also, other interesting options to be considered are turbocharging and direct injection system. These could improve the volumetric efficiency in high dilution conditions such as S50D and S75D at λ ≤ 1.4.
The stable operation of an internal combustion engine is one of the most important parameters, and it is a critical task when alternative gaseous fuels, derived from biomass waste, are applied [
30]. In this work, it was quantified with the COV
IMEP (
Figure 8) and a limit below 3% was taken as a reference of stable combustion processes [
36]. Pure and diluted syngas show excellent behavior in extreme lean conditions as well. The lack of spark timing optimization affected pure fuel more than the real one. Methane at λ = 1.4 presents values over the stability limit. This suggests the advantage of using hydrogen as an additive to control the combustion process in SI-ICE.
Fuel conversion efficiency is depicted in
Figure 9. It was calculated as defined by Heywood [
30], i.e., the ratio of the work produced per cycle (
) to the amount of fuel energy supplied per cycle that can be released in the combustion process:
Cyclic work was calculated based on in-cylinder pressure measurements, and the energy supplied was estimated by the fuel mass multiplied by the lower heating value (LHV) of the fuel (see
Table 3). The baseline fuel shows values of around 20% for all the air dilution tested. However, pure syngas presents an improvement when λ is higher. This is mainly due to the spark timing configuration, which is better for lean mixtures of these gases (S50 and S75). On the other hand, real syngas mixtures exhibit a decrease with air dilution for both hydrogen content. The losses in volumetric efficiency have greater penalties for these fuels (S50D, S75D) as was explained in the IMEP analysis. The highest efficiency values were seen for real syngas mixtures at λ = 1.0 (21.5%). It is important to note that the low-efficiency values are common in optical research engines due to the increased top land volume and high blow-by losses [
37].
The last set of parameters analyzed in this section are the exhaust emissions, which were measured with a Multigas 2030 spectrometer analyzer.
Table 5 shows pollutant emission (CO, NO
x, and CH
4) for λ = 1.4. Only this condition was considered for the sake of brevity of the manuscript and because it is the most common condition in energy production applications. From the table, it can be seen that the concentration of CO at the exhaust increases with that in the intake manifold (due to fuel content). The high top-land region volume and low fuel conversion efficiency that was seen in the previous graph are the main reasons for this behavior. Similar behavior was seen during the methane combustion process with high concentration of CH
4 at the exhaust. Despite the fact that real syngas features lower CO percentages in the fuel than the corresponding pure blends, more unburned fuel was found in the exhaust line. In this case, the drop-in fuel conversion efficiency is the main explanation (see
Figure 9).
NO
x emissions are one of the most important aspects considered by the regulations, due to the direct impact on local air pollution. Moreover, it is also one of the most controlled gases when hydrogen is used in combustion because of the tendency to increase the temperatures in the combustion chamber (see
Table 3). NO
2 and NO generation during the combustion process is strongly correlated with in-cylinder temperature [
38]. Therefore, roughly the same trend as that shown in
Figure 6 (high peak pressure correlates with high burned gas temperature) can be appreciated in the measurements of NO
x concentrations.
3.2. Optical Investigations
Traditional thermodynamic analysis gives global information on the evolution of combustion. However, for the study of new fuels, it is important to introduce additional measures for the detailed study of the process. In this work, direct visualization was performed through the window provided by the piston crown.
Figure 10 shows an image sequence of six different instances (from 5% to 60% of piston area) for stoichiometric air-fuel ratio and syngas blends with 50% H
2. A similar sequence can be seen in
Figure 11 for 75% H
2 blends. Methane data was duplicated for easier comparison. The crank angle values at the different flame area thresholds give indirect information of flame propagation speed. It is clear that pure syngas is much faster than diluted syngas and methane. Despite the fuel dilution, the hydrogen content of real syngas mixtures improves the combustion speed and shortens its duration by 4 CAD for S50D and 6 CAD for S75D with respect to methane. The high difference in terms of luminosity between diluted syngas and the other fuels is mainly because of the change in the f-number of the focal length as was mentioned in the optical measurement section. However, quantitative measurements such as flame speed, distortion, and displacement if its center in the combustion chamber can be measured with high precision [
23].
In the case of lean combustion, the same camera set up could be used due to lower flame intensity.
Figure 12 and
Figure 13 show the sequence of images for H
2/CO at 50/50% and 75/25% for the leanest case, respectively. Syngas without fuel dilution shows the highest emission intensity, followed by diluted syngas and methane. This behavior could be explained due to a different rate of heat released and local concentration of active chemical species. Martinez et al. [
20] showed that the hydrogen content in mixtures of methane and syngas increases flame luminosity. The same behavior can be seen for these fuels. In addition, it can be appreciated that the flame front thickness was greater for pure syngas compared to the diluted blends and methane. Pure syngas featured the fastest propagation and the lowest flame distortion at the same air-fuel ratio. Also, the high content of hydrogen reduces the displacement of the flame center. This promotes combustion with a more uniform process.
After a qualitative analysis of the flame front propagation, this study quantified the trends of the average flame area and propagation speed by applying the image processing procedure described in the previous section. Results reported in
Figure 14,
Figure 15, and
Figure 16 are related to the averaged values over 30 consecutive engine cycles. Pure syngas was the ‘fastest’ fuel due to the high content of hydrogen. In particular, the H
2/CO 75–25% blend at λ = 1.0 featured the highest speed value (32 m/s), 20 m/s faster than the baseline fuel in the same conditions (
Figure 14). It is important to note that the results have the same trend as evaluated by using in-cylinder pressure data (see
Table 4). Specifically, valuable results could be extracted even in the 0–5% MFB range, which is well recognized to be less accurate when using traditional thermodynamic analysis. In addition, S75D and S50D showed an increase of 5 m/s and 3 m/s with respect to methane, respectively. Therefore, the presence of hydrogen compensates for the inhibiting effect of diluent gases (CO
2 and N
2) on the flame propagation speed.
The lean cases (λ = 1.2 and λ = 1.4) showed a similar trend as λ = 1.0. For the leanest operating point, S75D suffered a decrease of 8.7 m/s with respect to λ = 1.0 and in the case of S50D, a decrease of 5.5 m/s was found (
Figure 17). This analysis can be performed thanks to the fixed spark timing setting; these trends would change if optimized ignition were to be set for each fuel. However, this point was not among the objectives of this work. Another interesting comment is the fact that these results contribute to the literature on flame propagation when using alternative gaseous fuels in ICEs. In addition, these trends can be included as a direct comparison for 1-D and 3-D numerical models.
As iterated in the previous sections, laminar flame speed values are commonly used as a reference point for evaluating combustion evolution in ICEs. The results presented in
Figure 17 for flame propagation speed inside the combustion chamber of the ICE (i.e., their peak values extracted from the traces shown in
Figure 14,
Figure 15 and
Figure 16) are in line with those of laminar flame speed at atmospheric condition in constant volume chamber devices (see
Table 3). This is associated with the fact that laminar flame speed is a key factor in flame front propagation. Also, the same fluid-dynamic conditions are set at ignition, and therefore, the turbulent component for different fuels is comparable [
39].
Measured flame propagation speed (turbulent speed) was also compared to laminar flame speed values calculated in the engine like conditions.
Figure 18a shows the comparison of the two parameters (i.e., their peak values, with the latter estimated using the CHEMKIN tool, with pressure and temperature values determined from measurements on the engine (at the crank angle of peak turbulent flame speed). It is important to note that pressure was directly taken from the pressure transducer. Meanwhile, the temperature was estimated with the first law of thermodynamics coupled with a heat and mass transfer model [
34]. The results show the same trend between fuels in laminar and turbulent flame speed. It is good to note that the graph was depicted with a scale factor of 10 between turbulent and laminar. Despite pure syngas having the highest values in both conditions, methane and diluted syngas presented high rates of increase, around 20 times with respect to 10 times of pure cases (
Figure 18b). This could be associated with the effect of turbulence in slow mixtures. Therefore, the results show that at the engine like conditions the values taken in constant volume chamber could be taken as a reference but turbulence influence need to be considered.
Proceeding with the morphologic analysis, flame distortion was analyzed by using the Heywood Circularity Factor (HCF).
Figure 19a shows that without air dilution (λ = 1.0) the flame propagated quite uniformly in all directions for pure and real syngas blends. Methane resulted in less “circular” flame fronts, with a maximum distortion of around 1.35. Focusing on the different relative air-fuel ratios (
Figure 19b), it is possible to see an increase at higher λ. This behavior could be attributed to the higher effect of in-cylinder turbulence due to the low burning velocity. However, hydrogen content helps to reduce this effect as can be seen for HCF values in the case of syngas with respect to methane. For real syngas mixtures, the double effect of air and fuel dilution caused a strong increase in flame distortion, even higher than pure methane.
Another global morphology parameter analyzed was flame center displacement. This parameter gives interesting information about how the flame evolves inside the combustion chamber and shows whether there is a preferential direction of propagation.
Figure 20 depicts the trends in stoichiometric conditions, for which an initial displacement towards intake valves can be noticed, after which a return to the center of the combustion chamber was measured. The first trend could be associated with the tumble motion (vertical plane) and the selected injection mode (PFI). For the same cylinder head used in this work and similar crank angle speed, Gomes et al. [
40] measured Tumble ratios of 4.0 during intake and 1.5 during compression. Centered flames during late combustion are expected because it is the center of the optical window (see
Figure 1b). Another important aspect is that almost negligible movement in the horizontal axis was seen for all fuels. For this type of combustion chamber geometries, the swirl motion (horizontal plane) is low. Huang et al. [
41] show that for a similar combustion chamber geometry the swirl ratio values are around 0.1 [
40].
Therefore, another comparison between fuels was done with maximum vertical displacement (
Figure 20b). It could be seen that the addition of hydrogen reduces the displacement towards the intake valves. However, the increase in fuel dilution enhances the preferential displacement to that side. In addition, air dilution was seen to have a similar effect to that of fuel dilution, e.g., during lean fueling with pure syngas (S50 and S75 at λ = 1.4) and when inert gas was added (S50D and S75D at λ = 1.0).
Finally, the flame front surface wrinkling was analyzed with respect to fuel type. This parameter is associated with in-cylinder turbulence and fuel chemistry [
38,
42]. The literature is not completely coherent on this aspect; however, several authors link flame front corrugation with the combustion reaction rate and overall duration of the process [
43]. Also, it was suggested that the high thermal and mass diffusivities of hydrogen reduce the flame thickness and with that increase wrinkling. These results are important for numerical simulation codes, as well, especially if wrinkling effects are considered when calculating the flame front area [
44]. In this work, as was described in the methodology section, wrinkling was measured with the curvature tool, as the average of 10 combustion images with approximately the same flame area for two different points (10% and 30% of the piston cross-section). This ensures a comparative study of all the fuels at similar MFB; the first threshold (i.e., 10%, equivalent to around 5% MFB) is representative for flame kernel development, and the second one (i.e., 30%, equivalent to around 10% MFB) is indicative of the fully turbulent propagation phase.
A flame curvature comparison between pure syngas, methane-hydrogen blends and pure methane [
20] revealed that higher hydrogen content increases flame wrinkling, as does the air-dilution. In this work, the analysis was focused on comparing real syngas mixtures (fuel diluted) and pure methane during lean fueling.
Figure 21a shows a common trend of a reduction in wrinkling for S50D, and then an increase for S75D with respect to methane; this was true for both flame sizes. To improve the analysis, HCF and S
L were also inserted (
Figure 21b,c, respectively). The graphs suggest that macroscopic flame distortion is related to flame distortion and less to the laminar flame speed. Several works [
45,
46,
47] suggest that higher flame speed reduces wrinkling, due to the fact that the flame front is less affected by the turbulence. Reiterating the discussion on flame-turbulence interaction, one important aspect that needs to be considered is that methane reaches the flame area thresholds later during the cycle compared to the two syngas blends, therefore in a crank angle region with different turbulence intensity. No clear affirmation can be stated with respect to higher or lower intensity near TDC (given the conversion of tumble to turbulence [
48]), but the common trend of increased wrinkling later during flame propagation could explain the relatively high value recorded for methane at 10% flame area compared to SD50. The fact that the latter fuel type features the lowest value for both instances emphasizes the importance of the second aspect of flame-turbulence interaction, i.e., combustion chemistry. Calculated laminar flame speeds do not seem to correlate with the observed trends of wrinkling.
On the other hand, flame-induced turbulence also needs to be considered. Even though it is still a matter of debate [
49], it is well established that there are basically three mechanisms determining wrinkling [
50]: eddy diffusion associated with turbulence in the unburned gases, which tends to increase it; propagation of the flame into the unburned region, which tends to reduce it; and instability, shear and eddy diffusion resulting from flame generated velocity gradients produced by the pressure drop across the combustion zone due to the density ratio, which tends to increase it. Considering these effects, higher laminar flame speed should be associated with less wrinkled flame fronts; on the other hand, it could increase the influence of the first mechanism, as previously iterated. The third effect should play a minor role, given that quite similar density ratios can be expected for all three fuel types. One definite observation is that the actual balance between the three mechanisms is what determines wrinkling trends, and a single explanation clarifies only part of the phenomenon. Looking at flame sensitivity from a more quantitative point of view, methane features an effective Lewis number around unity [
51]; therefore, turbulent flame speed is expected to be influenced to a relatively small degree by wrinkling. Changes in the values of this parameter can have a significant impact on stretched laminar flame speed, especially in the first stages of combustion [
52]. Even if values relatively close to unity should be expected for syngas as well [
53], slight modifications in the effective Lewis number could induce variations of laminar flame speed values. No clear conclusion could be drawn on the observed trends, other than the fact that, given that all cases featured quite close values, turbulence characteristics dominate over fuel chemistry effects with respect to flame wrinkling; nonetheless, an interesting line of future investigations was identified in the form of studying more in detail flame-turbulence interactions in syngas-fueled SI engines.