Chemiluminescence-Based Analysis of Syngas/Diesel Dual-Fuel Combustion in an Optically Accessible Engine

Abstract

Syngas (synthesis gas) is a promising gaseous biofuel for small-scale power generation, but its highly variable composition, which depends on the biomass source and gasification process, poses challenges for engine optimization. This study investigates syngas–diesel dual-fuel combustion in an optically accessible engine using chemiluminescence imaging of OH*, CH*, and CH₂O* to characterize ignition and flame development. Three representative syngas compositions—Downdraft, Updraft, and Fluidbed—were examined. The Fluidbed composition exhibited the weakest OH* signal, approximately one-third of that observed for the other two, primarily due to its higher CO₂ dilution and lower H₂ content. Ignition delay trends were strongly correlated with dilution level: Downdraft and Updraft showed similar delays despite different H₂/CO ratios, while larger CO₂ shares led to longer delays and flattened heat-release rates. CH* and CH₂O* chemiluminescence showed better agreement with combustion timing than OH*. Methane enrichment enhanced flame propagation and reduced ignition delay, partially offsetting CO₂ dilution effects.

1. Introduction

The energy sector is under significant pressure to transition to clean and renewable energy sources. To meet the ambitious goal of achieving net-zero carbon emissions by 2050, a target set by over 100 countries [1], all sectors of the economy must undergo substantial transformation. While many believe that a complete shift to a combustion-free energy sector is the only viable solution, achieving this is far from straightforward.

In this context, biofuels emerge as a promising alternative reducing global CO₂ emissions more affordably and with simpler technologies. One such biofuel is ‘syngas’, also known as producer gas or wood gas, which refers to gaseous fuel produced through Municipal Solid Waste (MSW) or biomass gasification. This process converts large biomass molecules into simpler gas molecules (H₂, CO, CO₂, etc.). Syngas, which is in essence a mixture of hydrogen (H₂) and carbon monoxide (CO), should not be confused with biogas, which is a mixture of methane (CH₄) and CO₂, produced from landfilling and methanation. Gasification, as a waste-to-energy process, competes with the two alternatives. While the carbon footprint of syngas is higher than that of landfilling [2], it can, in some cases, outperform biogas production [3]. However, from a Life Cycle Analysis (LCA) perspective, how efficient it is and how much Green-House Gas (GHG) is produced by waste after-treatment should be assessed on a case-by-case basis. As demonstrated by Dastjerdi et al. [4], gasification-pyrolysis could produce less GHG than landfilling, using MSW as feedstock.

The biomass type [5] and the gasification process itself [6] affect the syngas composition: a typical one includes H₂, CO and CH₄ with a volume concentration range of 10~20%, 13~24% and 0~7% respectively [6,7]. Additionally, air-fed gasifiers produce syngas compositions with 10~20% CO₂ and 45~60% N₂. In contrast, oxygen- or steam-based gasification permit a strong reduction in N₂ fraction, but often come with higher costs and are limited to specific use cases [7,8]. The three syngas compositions tested in this study replicate those of Bridgwater [6] and are from three types of gasifiers. Two are fixed-bed and one is a fluidized-bed gasifier, presenting different dilution and H₂/CO ratios, which influences the engine-relevant combustion characteristics, such as the laminar flame speeds [9] and the ignition delays [10].

Syngas is a biofuel that can be used in Combined Heat and Power units (CHP), that can be based on Internal Combustion Engines (ICEs) or on Gas Turbines (GTs). Since the auto-ignition temperature of syngas is rather high [11], a syngas-fueled ICE needs a source of ignition as an electrical spark or a pilot injection of a more reactive fuel [12]. The latter is usually known as a dual-fuel engine, and has the advantage of providing flexibility along with greater fuel-conversion efficiency [12]. The literature reports several studies on dual-fuel syngas ICEs, addressing engine efficiency and emissions across different syngas compositions [13,14,15,16,17]. Nevertheless, a deeper understanding of the combustion dynamics of the syngas dual-fuel operation is required to properly set the operating conditions of the engine, as a function of the syngas composition.

Optical diagnostics provide a direct window into the in-cylinder combustion processes, enabling spatially and temporally resolved characterization of ignition timing, flame propagation, soot formation, and flow-field dynamics. Among the most informative targets are chemically excited radical species whose emissions display distinct combustion regimes. In the present case, OH* marks the high-temperature reaction front, CH* delineates active flame zones, and CH₂O* is associated with low-temperature pre-ignition chemistry and cool-flame reactions [18,19,20].

Two established detection strategies are used to track these species. Natural chemiluminescence (CL) records the spontaneous photon emission from electronically excited radicals, requiring only a band-pass filter and an intensified camera. This technique is well-suited to optically accessible engines where access is geometrically constrained. Its main limitation is the line-of-sight integration, inherent to this technique, which precludes depth-resolved species quantification. In contrast, planar laser-induced fluorescence (PLIF) excites the target species with a tunable laser sheet, yielding two-dimensional concentration maps with potential for quantitative species measurements. However, this comes at the cost of substantially greater experimental complexity, as each target radical requires a dedicated laser wavelength and precise optical alignment.

A third approach was recently introduced by Mancaruso et al. [21], who demonstrated high-speed infrared imaging as a non-intrusive diagnostic capable of capturing qualitative fuel-vapor distributions and identifying ignition sites under both low- and high-temperature conditions. While this technique extends the diagnostic toolkit beyond the visible/UV spectrum, it requires a specialized high-speed IR camera, limiting its accessibility.

The combustion-emitted radiation in the visible and near-visible spectral range results from the superposition of two physically distinct contributions: soot incandescence and radical chemiluminescence. Soot incandescence produces a continuous broadband spectrum, with peak emission intensity governed by the solid particles’ temperatures. In contrast, radical chemiluminescence yields spectrally narrow emission bands generated when electronically excited species relax toward lower-energy states [19]. Radical chemiluminescence is a narrow-band emission, which is a result of the photons emitted during the electronic transitions from a self-excited to a more stable molecule or species [20], overlaid on top of the broadband emission from the CO₂* continuum [22]. In a dual-fuel configuration where a diesel-type pilot is delivered in minimal quantities, solely to initiate combustion of the premixed fuel, the overall natural luminosity becomes dominated by chemiluminescence (CL) rather than soot incandescence [23]. For syngas in particular, the H₂ content drives OH* as the principal radiative species, which differs from conventional hydrocarbon combustion behavior and its characteristic UV emission peak allowing spectral separation from CH* signals [24]. The OH* chemiluminescence signal exhibits its main intensity peak in the ultraviolet region, which allows it to be spectrally separated from CH* chemiluminescence. Nevertheless, Huang et al. [24] reported that increasing CO and/or CO₂ in the mixture can markedly attenuate the OH* intensity level. For syngas combustion initiated by a diesel-type pilot injection, the injection strategy can also promote the emergence of low-temperature pre-reaction chemistry prior to the main high-temperature heat release [25], mainly identified by excited formaldehyde radicals (CH₂O*), as observed in HCCI combustion [26]. In dual-fuel operation, monitoring OH* and CH₂O* can provide a practical means to separate the low-temperature (low-T) ignition stage from the subsequent high-temperature (high-T) ignition event. Still, for some mixture compositions and reactivity levels, the emissions associated with the pilot-fuel combustion, and those produced by the premixed syngas burn, can remain ambiguous [25].

A number of optical engine investigations have examined dual-fuel combustion with natural gas (NG) or methane initiated by a diesel pilot injection [27,28], and similar approaches have only been extended to ammonia-fueled dual-fuel operation more recently [29,30]. Syngas has not received the same level of attention in optical dual-fuel engine investigations to date. The limited optical studies available for syngas have largely been carried out in spark-ignition (SI) engines [31,32], where the emphasis is typically on fully premixed flame propagation and overall flame development rather than on imaging or tracking specific radicals and intermediate species.

Solferini de Carvalho et al. [32] examined premixed combustion of an SI optical engine, systematically enriching a representative syngas blend (H₂–CO–CH₄–CO₂–N₂), increasing the H₂ fraction from 14% (baseline composition) up to 62%. Their experiments were conducted in an optically accessible SI engine (CR = 9.68:1) equipped with an intensified high-speed camera to visualize and quantify flame evolution. They reported that turbulent flame speed rose approximately linearly with H₂ content, and that a mixture containing 24% H₂ exhibited a flame propagation speed comparable to that of neat CH₄. In addition, increasing H₂ led to smoother flame fronts, with a clear reduction in flame wrinkling. As noted in [33], the ratio between turbulent and laminar flame velocities suggests that, under H₂ enrichment, the effect of turbulence on flame acceleration is less significant than the influence of mass diffusivity and chemical reaction rate. These results are consistent with other optical SI engine studies on reformer gas mixtures (H₂–CO–CO₂) [33,34]. Martinez-Boggio et al. [35] also recently confirmed these results by analyzing natural luminescence images in the UV-visible spectrum along the entire cycle to follow the flame front development, for different compositions of H₂-CO-CO₂-N₂.

Hydrogen addition predictably enhances flame propagation and tends to stabilize combustion, yet the largest turbulent-to-laminar flame speed ratio is not necessarily obtained at the highest H₂ fraction, but can instead occur for mixtures containing larger CH₄ and/or diluent fractions. For typical syngas blends, the turbulent-to-laminar ratio is in the order of 20, highlighting how turbulence strongly amplifies a comparatively low laminar flame speed, whereas for undiluted H₂/CO mixtures the ratio is closer to 10 [32].

In an NG dual-fuel optical engine study, using methane as a surrogate, Ahmad et al. [27] investigated how the premixed methane/air equivalence ratio influences ignition delay, while also varying intake temperature and the diesel pilot energy share. At the lowest pilot contribution (16.5% of total fuel energy), soot incandescence “bright spots” were reported only late in the combustion event and remained localized near the injector tip, and increasing the pilot fraction from 16.5% to 27.5% produced only a small reduction in the measured ignition delay times (IDTs).

Srna et al. [36] examined chemiluminescence signals from lean premixed natural gas combustion initiated by a dodecane pilot in a rapid compression–expansion machine. They showed that the measured natural luminosity contains strong contributions from soot incandescence as well as OH* and CH* emission, with additional broadband interference associated with CO₂*, CHO*, and CH₂O* chemiluminescence. A notable outcome of their analysis was that CH* chemiluminescence is essentially confined to the onset of ignition and the interval dominated by pilot-fuel combustion. They further extended their work by applying CH₂O PLIF measurements [18]. Their results indicated that, depending on the diesel fraction, soot incandescence can contaminate natural luminosity-based interpretations, so that localized bright spots may still appear even when using OH* UV chemiluminescence detection around 308 nm [27,36]. Their goal was to assess why ignition delay times (IDTs) inferred from heat-release rate (HRR) analysis can differ from an IDT defined by the first detectable OH* chemiluminescence in dual-fuel operation. They reported that enriching the methane/air mixture and reducing the intake O₂ concentration both suppress the OH* signal and the broadband luminosity (dominated largely by CHO*, CH₂O*, and CO₂* emissions), which can render difficult a reliable identification of the true ignition onset. Cheng et al. [37] combined OH* and CH₂O* chemiluminescence measurements with broadband natural-luminosity imaging in a methane/diesel dual-fuel optical engine. Using CH₂O* CL to represent the low-temperature stage and OH* CL to mark high-temperature ignition, they reported only a small separation between the corresponding ignition delay times. Even so, the CH₂O* signal increased more rapidly than OH*, and they attributed this behavior to the short pilot injection duration, which reduced the temporal gap between the two combustion phases.

This paper presents and analyzes combustion in a syngas–diesel dual-fuel engine using chemiluminescence imaging for three syngas compositions, Downdraft, Updraft and Fluidbed (Fluidized Bed), as defined by Bridgwater [6]. First, OH* CL images and average signals are presented and discussed, followed by CH* and CH₂O* CL, highlighting the role of excited-state formaldehyde during low-temperature ignition and its consumption in the second combustion stage. The effect of varying CO₂ dilution and CH₄ enrichment, individually on H₂ + CO mixtures from the three compositions, are then assessed through changes in the averaged OH* signal to analyze its cycle-to-cycle variability.

To the best of the authors’ knowledge, this study is the first to use chemiluminescence imaging to systematically examine syngas–diesel dual-fuel combustion in an optically accessible CI engine for the representative syngas compositions defined by Bridgwater. The central question addressed here is how changes in syngas composition, particularly CO₂ dilution, H₂/CO balance, and CH₄ enrichment, alter ignition timing, radical development, and flame evolution under dual-fuel compression–ignition conditions.

2. Experimental Set-Up and Methodology

2.1. Experimental Set-Up

The experiments were performed on an optically accessible single-cylinder research engine derived from the PSA DW10 cylinder-head architecture. Key operating conditions and hardware specifications of the test engine are summarized in

Table 1. Engine characteristics.

Displaced Volume:	499 cm3
Bore:	85 mm
Stroke:	88 mm
Optical Window Diameter:	50 mm
Rod Length:	145 mm
Compression Ratio (CR):	15.6:1 *
Swirl Ratio:	2.0
Piston Bowl Type:	“Mexican hat”
Firing TDC Position:	0 CAD
Intake Valve Opening:	351 CAD ATDC
Intake Valve Closure:	157 CAD BTDC
Exhaust Valve Opening:	140 CAD ATDC
Exhaust Valve Closure:	366 CAD ATDC
Coolant Temperature:	85 °C (358 K)
Oil Temperature:	80 °C (353 K)

* Slightly lower CR than the standard engine.

.

The crankshaft speed was set to 750 rpm, a common speed of European gensets, and to limit thermal and mechanical loading of the piston-mounted optical window, combustion was initiated only once every three engine cycles. A schematic overview of the full experimental arrangement is provided in Figure 1.

The intake temperature was maintained at 25 °C (298 K) and the intake pressure at 1 bar. Temperatures were recorded at the intake and the exhaust of the engine, as well as the intake and exhaust pressures. The in-cylinder pressure traces were recorded by an AVL GH15D sensor and coded every 0.1 CAD. The post-processing method was described in [17], but from the 120 recorded in-cylinder pressure cycles, only 40 are fired cycles and 80 are non-fired cycles. The apparent Heat-Release Ratio (HRR) is plotted, instead of the combustion (net) HRR, because the calculation of thermal losses is less straightforward in an optical engine and could induce interpretation errors. The apparent HRR was obtained from the standard single-zone pressure–volume formulation using Equation (1), where $\gamma$ is the heat capacity of the gas in the cylinder.

$$\mathrm{H}\mathrm{R}\mathrm{R}={\displaystyle \frac{\mathrm{\gamma }}{\mathrm{\gamma }-1}}\mathrm{P}{\displaystyle \frac{\mathrm{d}\mathrm{V}}{\mathrm{d}\mathrm{\theta }}}+{\displaystyle \frac{1}{\mathrm{\gamma }-1}}\mathrm{V}{\displaystyle \frac{\mathrm{d}\mathrm{P}}{\mathrm{d}\mathrm{\theta }}}$$

(1)

Air is first compressed and dried before passing through the Brooks mass-flow controllers. The complete flow-metering configuration is summarized in

Table 2. Mass-flow controllers.

Gas Type	Controller	Full Scale	Uncertainty
AIR	Brooks 5853S	500 NL/min	±0.5%
N2	Brooks 5851S	100 NL/min	±0.9%
CO	Brooks 5851S	100 NL/min	±0.9%
H2	SLA5850	50 NL/min	±1.0%
CO2	SLA5850	37 NL/min	±1.0%
CH4	SLA5850	5 NL/min	±1.0%

and yields an overall uncertainty of ±2% on the intake–charge equivalence ratio. For the pilot fuel, decane (C₁₀H₂₂) is delivered using a Bosch CRI 2.2 common-rail injector operating at a reduced rail pressure of 200 bar, enabling stable and highly repeatable small injections. The injected mass at this reduced pressure was previously calibrated using an IAV type-K flow injector. The start of injection (SOI) was fixed at 15 CAD BTDC for all conditions, with an injection duration of 400 μs corresponding to the minimum pilot quantity required to ensure stable operation for the three literature syngas compositions [6]. Under these settings, the injected mass was approximately 0.32 mg, i.e., below 2% of the total fuel energy, corresponding to $\varphi$_decane/air $\cong$0.02.

2.2. Optical Imaging Set-Up

Chemiluminescence imaging was performed with a Phantom v1610 monochrome high-speed camera equipped with a Nikkor 105 mm objective and coupled to a high-speed image intensifier High-Speed IRO, from LaVision GmbH, Gottingen, Germany. As illustrated in Figure 2, the camera–intensifier assembly was positioned in line with a 45° mirror, and combustion luminosity was viewed through the extended piston and its fused-silica window, whose bowl geometry matches the original engine’s “Mexican hat” design. For OH* measurements, a FSQ-UG11 band-pass filter was combined with a 20CGA-305 long-pass filter (both provided by Micro-Controle Spectra-Physics, Beaune la Rolande, France) to favor transmission around 308 nm while suppressing out-of-band radiation. This OH* filter combination (330 ± 50 nm) does not block the contributions from other species, so in the IDT detection attention must be devoted to avoiding misinterpretations of the results. However, the small pilot injection and the substantial H₂ fraction in the premixed charge allow us to assure that the main participant in the filtered signal is from the OH* chemiluminescence.

The resulting optical layout, including the spacing between iris, lenses, and imaging components, is provided in Figure 2. The intensifier settings were fixed at 61% gain with a 39.2 ms gate and a 5 ns delay, while the camera was operated at 16,000 fps with a 45 μs exposure time, yielding three frames per crank-angle degree (CAD). Each acquisition started at 10 CAD BTDC and recorded 150 images per cycle at 768 × 768 pixels², corresponding to a magnification ratio of 11.5 pixels/mm. The high-speed image acquisitions were recorded for the 40 firing cycles.

The CH₂O* and CH* chemiluminescence signals [37,38] were acquired through a Image Doubler (from LaVision GmbH, Gottingen, Germany). CH* emission was isolated using a Chroma Technology band-pass filter centered at 431 nm with a 28 nm bandwidth, whereas CH₂O* (formaldehyde) chemiluminescence was captured using a stacked set of three bandpass filters (ZUL0325 LP, ZHS0385 SP, and ZUS0350 SP from Asahi Spectra USA, Inc., Torrance, CA, USA). It should be emphasized that the OH* filter combination does not fully reject CH₂O* signal; however, because CH₂O* chemiluminescence is typically orders of magnitude weaker than the OH* signal, its contribution is assumed negligible for the OH* measurements. For CH₂O* and CH* imaging, the intensifier gain was raised to 69% while keeping the same gate and delay settings, and the objective was replaced with a custom lens providing adjustable focal length, resulting in 768 × 576 pixels² images at 7 pixels/mm with the same exposure time. The combined-filter transmittance curves for each monitored radical are reported in Figure 2.

2.3. Imaging Post-Processing

Chemiluminescence data for OH*, CH*, and CH₂O* were post-processed using an in-house MATLAB R2024a workflow:

• First, a circular region-of-interest matching the piston-window diameter was applied to exclude parasitic light reflected by the piston crown and cylinder walls, so that only radiation transmitted through the quartz window was retained for analysis.
• To reduce background noise, a background image was generated for each cycle by averaging three images acquired 3 crank-angle degrees (CADs) before the start of combustion. This background was subtracted from each individual image. The corrected images were then ensemble-averaged at each CAD over 40 cycles, and the corresponding mean pixel intensity was computed.
• In addition, the ignition timing for each individual cycle was detected from a moving-median filter applied to a 10 × 10 pixel² interrogation window: when the median intensity in this window exceeded twice the median value of the same window over the preceding four CADs and was also above the noise threshold (100 counts), the corresponding CAD was flagged as an ignition spot.
• The ignition spot was then verified manually for each firing cycle to avoid misidentification. This was necessary for the CO₂’s diluted composition variations, where the ignition CAD could be mistakenly spotted.

2.4. Methodology

To isolate compositional effects in the syngas/diesel dual-fuel configuration, the baseline H₂/CO ratio of each reference syngas (Fluidbed, Updraft, and Downdraft) was preserved, while the N₂ fraction was adjusted to compensate for the added CO₂ and/or CH₄. The resulting gas mixtures are listed in

Table 3. Variations on the original syngas compositions. The original syngas compositions of Bridgwater [6] are shaded in blue, green and red for Fluidbed, Updraft, and Downdraft respectively. This set of colors is kept in the figures presented in the following sections.

Variation	H2 (%vol)	CO (%vol)	CO2 (%vol)	CH4 (%vol)	N2 (%vol)	Diluents- CO2+N2 (%)	LHV (KJ/m3)	Tad. (K)	SU° (cm/s)
FB–0CO2-CH4	9	14	0	0	77	77.0	2599	1868	32.3
FB–10CO2			10	0	67	77.0	2599	1834	24.3
FB–20CO2			20	0	57	77.0	2599	1802	14.9
FB–2.5CH4			0	2.5	74.5	74.5	3448	1955	33.0
FB–5CH4			0	5	72	72.0	4297	2016	35.0
FB–REFERENCE			20	7	50	70.0	4977	2004	26.4
UD–0CO2-CH4	11	24	0	0	65	65.0	4003	2092	61.5
UD–10CO2			10	0	55	65.0	4003	2059	51.1
UD–20CO2			20	0	45	65.0	4003	2028	41.9
UD–2.5CH4			0	2.5	62.5	62.5	4852	2131	59.5
UD–5CH4			0	5	60	60.0	5701	2160	58.2
UD–REFERENCE			9	3	53	62.0	5022	2112	51.1
DD–0CO2-CH4	17	21	0	0	62	62.0	4257	2120	83.0
DD–10CO2			10	0	52	62.0	4257	2088	69.1
DD–20CO2			20	0	42	62.0	4257	2057	57.9
DD–2.5CH4			0	2.5	59.5	59.5	5106	2154	75.4
DD–5CH4			0	5	57	57.0	5955	2178	71.2
DD–REFERENCE			13	1	48	61.0	4597	2095	62.7

. The original syngas compositions of Bridgwater [6] are shaded in blue, green and red for Fluidbed, Updraft, and Downdraft respectively. This set of colors is kept in the figures presented in the following sections. Following the operating strategy established in a previous study [17] on a comparable metal engine, the premixed syngas/air equivalence ratio was fixed at 0.7 and the pilot quantity of the pilot fuel was reduced to the minimum level required for stable operation. Keeping the pilot injection as small as possible helps suppress soot incandescence contributions to the optical signal, which otherwise appear as localized bright spots in natural-luminosity imaging [27].

3. Results and Discussion

3.1. The Case of Decane Pilot Ignition by Itself

OH* images are first recorded with the engine operating on decane pilot injection only. As shown in Figure 3a, from around 8 CAD BTDC to 10 CAD ATDC, ignition develops along the six injector plumes, with noticeable differences among jets due to hole-to-hole irregularities. Because the injected decane mass is very small, the images reveal only six compact flame kernels rather than fully developed ones, without the bowl-filling flame as in conventional diesel operation. The maximum area of natural chemiluminescence strongly depends on the ignition location, which in this configuration follows the decane spray. The overall burn duration remains below 14 CAD.

Using these data, the spatially averaged OH* chemiluminescence signal is compared with the apparent HRR derived from the in-cylinder pressure (Figure 3b), and the two traces exhibit good qualitative agreement. However, the small pilot quantity also renders the pressure-based HRR estimate particularly noisy, resulting in a low signal-to-noise ratio. The ignition delay time (IDT), defined as the crank-angle interval between SOI and the OH* signal reaching 1% of its maximum, is approximately 7 CAD, which is consistent with the HRR-based estimate. As expected, OH* chemiluminescence persists after the HRR peak. The emission detected near 310 nm may arise either from OH* chemiluminescence or from the CO continuum [39] associated with CO-to-CO₂ oxidation, and because these processes occur concurrently, the dominant contribution at 310 nm cannot be unambiguously identified.

3.2. Comparison of the Different Syngas Compositions

OH* chemiluminescence cycle-averaged images for the three syngas compositions are displayed in Figure 4. Downdraft and Updraft exhibit very similar combustion evolution and OH* intensity levels (Figure 4a,b): ignition still initiates along the decane spray, and even as the flame propagates further into the bowl volume, the characteristic multi-jet structure remains visible. Relative to operation with decane pilot injection only, the onset of OH* chemiluminescence shifts later by about 5 CAD for both Downdraft and Updraft, and by roughly 6 CAD for Fluidbed (Figure 4c). In addition, the Fluidbed case displays a lower OH* chemiluminescence level, which is consistent with two key compositional differences, namely the H₂ fraction and the overall dilution provided by CO₂ + N₂.

Cycle-to-cycle variations in IDT, based on the OH* onset, are presented in Figure 5 for each syngas composition, alongside its average and standard deviation in CAD. It is evident that the presence of CO₂ is one key parameter for the ignition delay, as the Updraft composition, with the lowest amount of CO₂ (9%), is the first to be ignited, followed by Downdraft and Fluidbed. The 20% CO₂ share in the Fluidbed composition also influences the cycle-to-cycle variations with the largest value, whereas Downdraft and Updraft are much closer.

Figure 6 shows the evolution of the cycle averaged OH* chemiluminescence signal for the three syngas blends and for the decane pilot alone over the course of combustion. The presence of syngas increases the ignition delay time while reducing the overall OH* intensity. For the Fluidbed mixture, the OH* chemiluminescence level is roughly three times lower than the other two, primarily because of stronger dilution and its reduced H₂ fraction. Interestingly, the OH* peaks earlier for Fluidbed than for the other two, which is counterintuitive given its lower laminar flame speed. The influence of the CO₂ dilution will be discussed in the following sections.

Figure 7 provides instantaneous OH* chemiluminescence snapshots at several crank-angle positions within the same single engine cycle for each fuel composition, to emphasize the spatial features of the ignition process. As in the pilot ignition on air, without syngas in the charge, ignition does not occur simultaneously across the six decane sprays. Ignition generally initiates away from the jet tip, with the Fluidbed case constituting the main exception. Once OH* emission becomes sufficiently strong in localized regions at a given CAD, the signal progressively appears more spatially uniform throughout the chamber as the flame develops. Finally, these observations correspond to a single representative cycle, and since cycle-to-cycle variability is substantial, we cannot infer the IDT from Figure 7.

The spatially averaged OH* chemiluminescence (CL) intensity is compared with the ensemble-averaged apparent HRR in Figure 8. As expected, the Fluidbed mixture exhibits substantially weaker OH* signals, because of its higher CO₂ concentration. In addition, the OH* intensity peak does not occur at the same crank angle as the apparent HRR peak, consistent with observations reported by Srna et al. [36] and Schlatter et al. [40] when they compared ignition delay times derived from conventional HRR analysis (or from Schlieren imaging) with those inferred from the onset of OH* chemiluminescence in NG/diesel dual-fuel experiments performed in rapid compression–expansion machines. For the Fluidbed case, the OH* trace is particularly affected because the signal remains weak under strong CO₂ dilution (20% vol.), and its maximum appears earlier than the HRR peak. Similar behaviors were observed by Bihari et al. [41], whose OH* CL peaks were retarded with increasing EGR (dilution level) until the jump from 8% to 13.8%, where the luminosity value decreased but the peak temporal location (i.e., CAD) was, in fact, advanced.

To assess whether other chemiluminescent radicals track the apparent HRR more closely in a syngas/diesel dual-fuel engine, the same comparison is performed for CH* and CH₂O* signals, as shown in Figure 9. For all three syngas compositions, the cycle-averaged CH* and CH₂O* intensities match the HRR profiles more closely than OH*, both in ignition timing and in the location of the peak, which is in agreement with the findings of Cheng et al. [37] for an NG/diesel dual-fuel engine. CH₂O* chemiluminescence appears ahead of OH* and increases more rapidly. The lack of a clearly separate two-stage ignition, in either the HRR trace or the averaged chemiluminescence signals, is consistent with the very small pilot-fuel quantity, which shortens the induction period between the two stages [37]. The CH* signal emerges shortly after CH₂O* but rises faster. For the Fluidbed mixture, the OH* and CH₂O* peaks occur slightly before the HRR peak, whereas CH* remains better phased than the two other radicals. The differences in peak phasing of the compositions are due to the CO₂ fraction of the Fluidbed composition, which quenches the radical emissions.

3.3. Effect of the CO₂ and CH₄ Content on the Syngas Ignition

Since syngas composition varies widely, in CH₄ (a reactive component) and CO₂ (a diluent) contents, it strongly influences ignition delay times (IDTs) and flame development in dual-fuel combustion. As an example, Rajasegar et al. [42] examined n-heptane ignition delays as a function of CH₄ or H₂ addition, combining experiments with kinetic simulations. They reported a marked inhibition of n-heptane autoignition chemistry in the presence of natural gas, manifested by longer ignition delays and a shift to first-stage ignition, which depends on the premixed equivalence ratio. This inhibition results from several coupled effects, including reduced oxygen availability due to premixing, changes in mixture heat capacity, and the role of CH₄ as a radical sink.

Regarding CO₂, changes in its concentration mainly modify the dilution level. Increasing CO₂ also raises the mixture-specific heat by about 30%, which lowers the theoretical adiabatic flame temperature (T_ad., in Table 3) and reduces the laminar flame speed. In addition to these thermal effects, CO₂ also exerts important kinetic influences through third-body collisions and radical recombination pathways, which can slow chain-branching reactions and further suppress reactivity. The literature studies therefore indicate that CO₂ dilution exerts a stronger influence than N₂ dilution. In contrast, variations in CH₄ content primarily alter the overall reactivity of the syngas. CH₄ acts as an additional fuel component that can promote radical production and accelerate ignition, but it may also behave as a radical sink under some conditions by consuming OH* and other active radicals during its oxidation pathways. In the following subsections, the effect of the individual fractions of CO₂ and CH₄ on the syngas mixtures will be examined by following OH* chemiluminescence imaging and average signals.

3.3.1. The CO₂ Concentration Effect

As an illustration of the CO₂ effect, Figure 10 presents ensemble averaged OH* images for the Updraft composition only, since similar trends are observed for the other two. As the CO₂ fraction in the total diluent mixture (CO₂ + N₂) increases, the combustion onset shifts noticeably later and the overall OH* intensity decreases substantially. At 20% CO₂, the emission becomes very weak and is largely undetectable without rescaling the images. This is further illustrated by the average OH* chemiluminescence signals plotted in Figure 11, where the Downdraft composition, which normally yields the highest OH* intensity, is strongly affected by CO₂ content. Replacing 10% of the fuel mixture of N₂ with CO₂ reduces the OH* CL signal to a level comparable to that of the Updraft case without CO₂, albeit with a slightly longer ignition delay. The IDT difference between these two syngas compositions is therefore primarily attributable to the higher CO₂ content of the Downdraft mixture.

Figure 11 presents the effect of CO₂ on the raw average OH* chemiluminescence signal (a) and a comparison with the apparent HRR, grouped by base composition (b). It must be noted that the recorded images of the Fluidbed composition without CO₂ or CH₄ are noisy, so additional post-processing is required. Compositions with 10% CO₂ yield a higher intensity, while the phasing of the OH* signal remains unchanged. The lower CO₂ content of the Updraft composition accounts for the marginally longer IDT of the original Downdraft composition, as visible in Figure 8. Regarding the HRR on Figure 11b, we can see the presence of CO₂ instead of N₂ strongly delays and reduces (i.e., ‘flattens’) the apparent HRR, especially in the Fluidbed case where, as an example, the IDT is delayed by 4 CAD from 0 to 10% of CO₂ and up to 8–9 CAD for 20% CO₂.

The same cycle-to-cycle variability analysis as in Figure 5 is performed, and the results are presented in Figure 12. Just as with the average ignition delay, CO₂ dilution increases the cycle-to-cycle variability in the engine: the standard deviation of the ignition timing rises from 0.41 to 0.60 CAD when CO₂ replaces 20% of the nitrogen content. This tendency is consistent with the findings of Gupta et al. [43], who demonstrated the same effect when analyzing the CO₂ dilution ratio in biogas compositions.

3.3.2. The CH₄ Concentration Effect

The effect of CH₄ is also studied, but only at 2.5 and 5% vol. As shown in Figure 13, OH* is strongly enhanced by the presence of CH₄ for Updraft, and similar trends were observed with the other two syngases. Figure 14, through the evolution of the average OH* chemiluminescence signal compared with the apparent HRR, shows how CH₄ addition induces a better combustion phasing, most noticeably in the case of the Fluidbed composition.

Figure 15 highlights that, as the CH₄ concentration increases, combustion is slightly advanced, accompanied by a rise in the HRR peak. The ignition advance can be quantified using the CAD value at which the HRR reaches 1 J/CAD, referenced to the SOI: a 2.5% replacement of N₂ with CH₄ advances ignition by 0.4 CAD, while a 5% replacement advances it by 0.7 CAD.

The zoomed portion of the graph highlights the small first phase of the two-stage combustion, which results from the small pilot-fuel injection as discussed in [17]. Nevertheless, the HRR results from the optical engine make it easier to distinguish between the two combustion phases.

Figure 16 illustrates the influence of CO₂ and CH₄ concentration on the ignition delay, represented by the CA10–SOI crank-angle value, and on the first combustion phase (CA50–CA10). As expected, increasing CO₂ content, which slows down the chemical kinetics, progressively delays ignition across all three syngas compositions. Conversely, CH₄ addition slightly advances ignition, an effect that is especially notable for the original Fluidbed composition, where the reactive methane shares partially offset the inhibitory influence of CO₂ and enhances the overall mixture reactivity.

The lower half of Figure 16 presents the duration of the first half of combustion (CA50–CA10) for the different syngas mixtures. The two additions behave differently for the Fluidbed-based composition compared to the other two: in the Fluidbed case, CO₂ addition doubles the CA50–CA10 when 20% of the mixture is changed from N₂ to CO₂, a trend that is consistent with the predicted reduction in laminar flame speed (32.3 → 14.9 cm/s). These results confirm that while CO₂ addition deteriorates combustion reactivity, moderate CH₄ enrichment can mitigate ignition delays in dual-fuel syngas operation. For the Downdraft and Updraft compositions, however, the reduction in laminar flame speed—from 83 to 71.2 cm/s for Downdraft and from 61.5 to 58.2 cm/s for Updraft—does not translate into a longer combustion duration in the engine.

4. Conclusions

To improve the knowledge about the effect of pilot-fuel injection as a function of syngas composition, chemiluminescence of OH*, CH* and CH₂O* images were compared to the heat-release rate in the case of the syngas/diesel dual-fuel combustion mode engine. Three syngas compositions, selected as representative of different gasification processes, are examined, and the effects of CO₂ and CH₄ content are investigated. The main conclusions are as follows:

• Under the selected syngas/decane dual-fuel operating conditions, no soot incandescence is detected in the OH* and CH₂O* chemiluminescence signals, which is attributed to the small, injected mass of decane. However, soot interference is noted in the CH* chemiluminescence images (i.e., at 431 nm), manifesting as a plateau in the corresponding average signal.
• Updraft and Downdraft display comparable OH* behavior, with similar ignition delays and overall intensity levels. The Fluidbed blend behaves differently, showing an OH* intensity about three times lower and a longer ignition delay. Increasing CO₂ dilution weakens the OH* maximum and, when a certain limit is reached, OH* chemiluminescence can peak earlier than the Heat-Release Ratio’s (HRR).
• When the OH* average signal is compared with the apparent HRR, their peak timings do not coincide, indicating that OH* is not perfectly phased with the global heat-release evolution. The cycle-averaged CH* and CH₂O* signals follow the HRR profile more closely, although they still do not fully resolve the very beginning of combustion development.
• The impacts of CO₂ dilution and CH₄ enrichment are examined to rationalize the ignition delay trends measured in the engine. On the Fluidbed blend, the higher CH₄ fraction counterbalances its larger CO₂ share, leading to comparable ignition behavior despite the stronger dilution. In contrast, Updraft gas combines a higher CO content with a lower CO₂ level, which can plausibly explain why its ignition delay remains close to that of the Downdraft mixture, in line with reports that variations in the H₂/CO ratio have no major effect on ignition delay.

Future work could expand these findings by focusing on varying parameters that were kept fixed in this study, like the pilot-fuel injection parameters and the premixed charge syngas equivalence ratio, in an approach following the Design of Experiments (DoE) method. A dedicated sensitivity analysis of the ignition delay time (IDT) with respect to individual syngas components would be highly valuable. Such an analysis would quantify the relative contribution of each constituent to the overall ignition behavior. For instance, while the present study demonstrates that CO₂ dilution exerts a stronger suppressive effect on ignition than N₂ dilution, and that CH₄ enrichment partially offsets this penalty, the precise sensitivity of IDTs to incremental changes in each component remains to be quantified under engine-relevant conditions. A response-surface or factorial DoE approach would enable such a decomposition while controlling for cross-coupling effects between composition variables. This would provide a more predictive framework for optimizing dual-fuel engine operation across the full range of syngas compositions produced in real biomass gasification systems.

These results, presented in this paper, constituted the final chapter of the first author’s PhD thesis manuscript [44].