Prechamber Equipped Laser Ignition for Improved Performance in Natural Gas Engines

Lean-burn combustion is highly preferred in stationary natural gas engines as it offers high-efficiency with concomitant low-emissions without the need for using an aftertreatment system. However, due to the fact that most of these engines are turbocharged, ignition is compromised as the lean mixtures are under sufficiently high pressures at the time of ignition. The situation demands spark gap voltages in excess of 40 kV, which standard electrical ignition systems cannot reliably transmit. A potential pathway to overcome this limitation while extending the lean ignition limit has been to use laser ignition.

where CR is the compression ratio, and γ is the ratio of specific heats. Taira et al. [1] identified the primary mechanism for efficiency improvement with lean mixtures to result from increased γ values.

However, the flame velocities decrease significantly in lean mixtures: The laminar burning velocity values calculated by using chemkin pro and gri 3.0 chemical kinetic mechanism for typical pressure and temperature conditions at the time of ignition are shown in Fig. 1. One notices that the laminar burning velocity strongly decreases as the mixture becomes leaner, for example, the laminar burning velocity at λ = 1.75 is about 40% lower as compared to similar values at λ = 1.54. The actual combustion rate, however, is influenced by local turbulence and mixing in addition to laminar burning velocity. The decreased combustion rate offsets any potential gains in efficiency due to increased γ values. To compensate for this effect, one could use a prechamber in tandem with laser ignition to promote volumetric ignition under extremely lean conditions.

In an unscavenged prechamber (see Fig. 2), residual exhaust gases from previous combustion cycle are displaced by fresh combustible mixture during the upward compression stroke. Thereafter, when an ignition kernel is created, it ignites the mixture within the prechamber and the pressure is locally increased. Jets of radical-rich partially combusted gases issue from the prechamber nozzle holes, each of which acts as an ignition source for the lean mixture in the main combustion chamber. The advantages include

• (i)

  Volumetric ignition—spatial distribution of the jets leads to faster combustion in the main combustion chamber.

• (ii)

  In-cylinder turbulence enhancement—each of the partially combusted jets issues from the prechamber nozzles at very high velocities (Reynolds numbers in excess of 30,000) and in turn enhances the turbulence within the main combustion chamber [2].

• (iii)

  Ignition energy amplification—depending upon the amount of the fuel–air mixture trapped in the prechamber at the time of ignition, amplification of ignition energy occurs up to 300 fold.

The objective of this study was to evaluate the efficacy of prechamber when used in tandem with laser ignition. First, the prechamber geometry was optimized through several tests in a single-cylinder engine. Subsequently, tests were performed comparing the performance of the three ignition systems: SI, LI, and PCLI.

A single-cylinder natural gas engine (RSi-130Q) in Argonne National Laboratory was utilized in this study. Engine specification and schematic representation are shown in Table 1 and Fig. 3, respectively. It is known that the single-cylinder engine suffers from high cyclic variation due to the pulsation in the intake and exhaust manifolds. Two surge (237 liters each) tanks are employed to dampen the flow pulsation. Also, with the help of two electronic valves, the surge tanks were used to simulate the turbocharger conditions. An intake pressure in the range of 5–24 psi gauge was used to achieve the required brake mean effective pressure (BMEP) for each air to fuel ratio (λ 1.54, 1.58, 1.65, 1.68, 1.7, and 1.75). The engine is coupled to a 111 kW AC dynamometer to facilitate ignition testing. Also, external lubrication oil and coolant systems were employed to supply the required lubricant oil at 85 °C and engine coolant at 90 °C, respectively.

Pipeline natural gas (PLNG) was used during the experiments. The composition of the natural gas changes slightly on a daily basis, however, a gas chromatography analysis was carried out to evaluate the composition over several days. The stoichiometric air to fuel ratio was found to be 16.39; gas composition for one of the tests is presented in Table 2. The natural gas was compressed from 5 to 130 psi gauge by using a CompAire natural gas compressor before directing it to the injector block. The PLNG flow measurement was obtained by utilizing a Micro Motion Coriolis flow meter downstream of the compressor. The fuel was injected into the intake manifold with the help of two electronically controlled natural gas injectors (CAP Inc.). Horiba MEXA-7100D emission bench was used to measure the composition of exhaust gases (CO, CO₂, NO_x, UHC, and O₂). In each test case, average values of data were recorded over 3 min and used for subsequent analysis.

The in-cylinder pressure was recorded by using a piezoelectric pressure transducer (Kistler 6013B) mounted on the cylinder head. Also, a charge amplifier (Kistler 5010) was employed to convert the charge into a voltage signal. The transducer records at 100 kHz which gives approximately ten data points every crank angle at 1800 RPM. Kistler 2614A optical shaft encoder was coupled with the engine crankshaft to determine the exact piston location. The data were recorded with a high-speed data-acquisition system (avl indicom) to perform the thermodynamics analysis. Five hundred consecutive cycles were recorded for each test condition; to minimize the cyclic variation, an average pressure is used to perform the heat release analysis. avl concerto software was used for the pressure data analysis.

The three igniters used in this study are schematically shown in Fig. 2.

Spark Ignition (SI): A standard capacitance discharge ignition system (Altronic CD200) was used in tandem with a standard 18 mm J-style spark plug (Altronic L1863ip). The system was capable of generating up to 30 kV pulses across the gap with an average energy of 35 mJ/strike.

Laser Ignition (LI): The laser igniter comprises of a microlaser affixed to one end of a hollow tube. The other end carries a sapphire lens (back focal length = 8.6 mm) with appropriate sealing to prevent the combustion gases from entering the hollow tube. The pulsed output from the microlaser (two consecutive pulses, ∼15 mJ/p each, approximately 150 μs separation, 5 ns full width at half maximum (FWHM), and 1064 nm) when focused creates sparks inside the combustion chamber. Further details of the laser igniter are given in our previous publication [3].

Prechamber Equipped Laser Ignition (PCLI): The PCLI resembled the laser igniter described above in all design aspects, except that a prechamber was affixed to its distal end.

As previous studies by Roethlisberger and Favrat [4,5] show that prechamber geometry—i.e., prechamber volume, number of nozzle holes, nozzle diameters, and their placement—has a profound impact on ignition performance and the formation of NO_x, CO, and unburnt hydrocarbons. Using the basic frame work laid out by these researchers, optimization of prechamber that can be used with LI entails

• (i)

  avoiding soot formation within the prechamber,

• (ii)

  minimizing NO_x formation both in the prechamber as well as in the jets, and

• (iii)

  avoiding autoignition within the prechamber [6]—occurrence of one such event in the prechamber imparts a fast rising pressure pulse that often leads to failure of the window/lens material.

In our efforts, the prechamber volume was kept constant at 1600 mm³ corresponding to 0.08% of the combustion chamber volume, and three different designs were evaluated:

• D1—(three holes 2.0 mm dia + three holes 1.0 mm dia); staggered configuration,

• D2—three holes 1.6 mm dia, and

• D3—three holes 2.23 mm dia.

Designs D1 and D3 have the same total nozzle cross-sectional area. Designs D2 and D3 have the same configuration, except for the fact that hole sizes were different. Ignition delay and combustion durations for the three geometries for λ = 1.58 are shown in Fig. 4. As seen, the combustion durations are somewhat similar. Additionally, in D2, which has the smaller total nozzle cross-sectional area, autoignition was more pronounced and the design was discarded. Subsequent tests comparing D1 and D3 (see Fig. 5) showed that both ignition delay and combustion duration were lower with D1. Additionally, there was no incidence of autoignition in the case of D1. The superior performance of D1 is attributable to the fact that it uses 2 mm dia and 1 mm dia holes in a staggered fashion, which in addition to the spatial separation leads to temporally separated issuance of jets from the larger and smaller holes.

Overall, design D1—prechamber volume 1600 mm³, three holes 2 mm dia, and three holes 1 mm dia at 45 deg to the igniter axis—was the optimal geometry that would offer low NO_x emissions and at the same time avoid autoignition and soot formation. This prechamber geometry was used in PCLI for subsequent tests comparing the three ignition systems.

As is typical in reciprocating engines, for advanced ignition timing (IT), engine thermal efficiency increases but the NO_x emissions increase as well. On the other hand, for retarded ignition timing, efficiency and NO_x emissions decrease. Additionally, with ignition retard the ignition stability decreases (i.e., COV_IMEP increases), and in general, with ignition advance, the ignition stability increases (COV_IMEP decreases). For all of the test conditions used here, engine knocking was never encountered.

With the above trends in mind, engine tests were performed at a fixed speed of 1800 rpm and a load of 10 bar BMEP. For gradually decreasing values of λ, ignition timing (IT) sweeps were performed. Other researchers [7], especially those working with gasoline engines, have used optimal combustion phasing to correspond to CA50 coinciding with a fixed value of 5 deg ATDC. However, to accurately identify the optimal ignition timing for the slow burning natural gas fuel that was used here, for a given λ, IT was varied between ignition advance corresponding to the EPA emissions regulation (BSNO_x < 1.34 g/kWh) and ignition retard corresponding to the limit for ignition stability (COV_IMEP < 5%) as accepted in the industry.

For the purposes of discussion, we will define “ignition delay” to correspond to the time period between ignition timing and that corresponding to 10% mass burn fraction. In the case of SI and LI, this definition somewhat coincides with Sjoberg and Zeng's “inflammation time” [8], wherein the flame kernel survives its nascent laminar state and transitions to a fully developed turbulent deflagration. During this time scale, especially in lean mixtures, the stochastic variations in temperature, velocity, and turbulence could influence the fragile flame kernel to result in significant cycle-to-cycle variations.

In a similar manner, we will define “combustion duration” to correspond to the time interval between 10% and 90% mass burn fractions. This time period corresponds to most of the fuel chemical energy being converted to mechanical energy in the form of high-pressure combustion gases. A short combustion duration that is phased appropriately results in a major fraction of the heat release to occur at the top dead center, i.e., leading to constant volume combustion, which in turn leads to higher thermal efficiency.

The pressure curves for the three ignition systems, for two different test conditions (IT 13 at λ = 1.54 and IT 17 at λ = 1.58), are shown in Figs. 6(a) and 6(c), respectively. One notices that in both conditions, LI leads to faster combustion that results in a higher peak pressure value as compared to SI. As evident from the rates of heat release shown in Figs. 6(b) and 6(d), LI also leads to shorter ignition delay. This may be attributed to larger flame kernel developed in the case of LI, which gets further amplified by heat released from combustion gases entrained from the surroundings [9]. The net effect is accelerated combustion leading to higher peak pressures.

However, in the case of PCLI, the partially combusted jets issuing from the prechamber introduce spatially distributed ignition sites at multiple locations, especially at the jet head vortices [10], leading to earlier and much faster combustion. This results in peak cylinder pressures approximately 20 bar larger than in SI or LI.

These trends are further evident when one compares ignition delays (see Fig. 7) and combustion durations (see Fig. 8) for the three ignition systems: Ignition delays for PCLI are ∼10 deg CAD shorter than those corresponding to SI, and combustion durations are reduced on an average by 20 deg CAD in the case of PCLI. Such trends warrant delayed ignition phasing in the case of LI and PCLI for engine optimal operation.

The brake specific NO_x emissions (g/kW-hr) versus efficiency and the corresponding COV_IMEP versus efficiency plots for the three ignition systems are shown in Fig. 9. The allowable levels for BSNO_x per EPA regulation (1.34 g/kWh) and the industry acceptable value for ignition stability (COV_IMEP = 5%) are also shown in this figure with red horizontal lines. One notices that for all the three ignition systems, for a given λ value, NO_x emissions increase with increase in efficiency (resulting from ignition advance), whereas COV_IMEP decreases (i.e., ignition stability improves) at higher efficiency. Additionally, LI extends lean operation to λ = 1.65, whereas PCLI extends it even further to λ = 1.7. As a result, the optimal operational point for a given ignition system was chosen as the λ and IT combination that offers the maximum efficiency. This results in LI exhibiting an incremental efficiency improvement of Δη = 1.3%, whereas PCLI performs even better with an improvement of Δη = 2.1%. As mentioned before, this is largely attributed to the spatially distributed ignition sites facilitated by PCLI, which lead to faster combustion and close to the top dead center. Additionally, the brake specific carbon monoxide and hydrocarbon emission values for the optimal operational points were close to one another, implying similar combustion efficiencies.

The main objective of this study was to evaluate the efficacy of using a prechamber in offsetting the reduction in efficiency that results from lower flame velocities under extreme lean-burn conditions facilitated by laser ignition. Toward this direction, through a series of systematic tests performed on a single-cylinder natural gas engine, prechamber geometry was optimized for use with laser ignition. In the process, using a unique prechamber geometry, the primary concern for integrity of laser delivery optics, i.e., autoignition within the prechamber, was avoided. Subsequently, tests were conducted comparing the performance of three ignition systems: spark ignition (SI), single-point laser ignition (LI), and PCLI. Out of the three, PCLI proved to have the best performance: It not only led to extension of the lean ignition limit but also shortened ignition delay and combustion duration significantly. As a result, within the bounds of EPA emissions limits and industry accepted ignition instability limits, an overall efficiency improvement of 2.1% points was observed. In summary, the larger initial flame kernel facilitated by laser ignition leads to extension of the lean ignition limit. Multiple partially combusted turbulent flame jets issuing from the prechamber, on account of their spatial distribution, lead to volumetric and faster ignition. As a result, overall ignition stability is improved under lean-burn conditions.

The authors gratefully acknowledge the help and support of various staff at Argonne and Princeton Optronics, Inc. Additionally, we acknowledge the partial support of UCF researchers by Public Authority for Applied Education and Training (Kuwait) and Argonne National Lab (Contract No. 6F-30762).

This work was performed under the financial support from the U.S. Department of Energy (DOE), Office of Energy Efficiency and Renewable Energy under Contract No. DE-AC02-06CH11357.

ATDC =

after top dead center

BMEP =

brake mean effective pressure

BSNO_x =

brake specific NO_x emissions (g/kW-hr)

COV_IMEP =

coefficient of variation of integrated mean effective pressure, %

IMEP =

integrated mean effective pressure

IT =

ignition timing (CAD BTDC)

MFB =

mass fraction burned

λ =

excess air ratio = (1/equivalence ratio)