Partially Premixed Combustion Application for Diesel Power Improvement

Military platforms continue to increase in weight due to technological and personnel safety requirements. In order to maintain mobility and performance, engine-specific power must be increased. However, diesel engine power is limited by soot formation, an indicator of incomplete fuel combustion, due to poor mixing of the air and fuel at high engine loads. More fuel cannot be delivered and reacted in the combustion chamber since the allotted diffusion time is very small at high revolutions per minute (rpm) and oxygen is limited at high engine loads. An alternative advanced combustion mode called partially premixed combustion (PPC) delivers a portion of the fuel to the cylinder via the engine air intake system, leading to the opportunity for increased specific power.

The objective of this study was to apply a PPC approach and the resulting reduction in engine-out particulate matter (soot) to extend the high load limit of a single cylinder diesel research engine based on exhaust opacity. Lim et al., in a principally modeling-based study, optimized a dual-fuel, dual-injection reactivity controlled compression ignition (RCCI) strategy of iso-octane then normal heptane to achieve 21 bar indicated mean effective pressure (IMEP) operation with low engine-out oxides of nitrogen (NO_x), soot, and carbon monoxide (CO) emissions [1]. Using the same fuels in a similar RCCI model, Klos and Kokjohn found that injection timing had significant tradeoffs between engine efficiency, emissions, and combustion instability [2]. A near top dead center injection timing significantly reduced combustion instability, while the emissions and efficiency dropped to near conventional direct injection levels. More advanced phasing was even more stable, but produced high peak pressure rise rates, more NO_x, and lower efficiency due to increased heat transfer losses.

Manente et al. used a single-fuel, PPC approach at 11.5 bar IMEP and found that high octane number fuels, like ethanol and gasoline, were able to achieve a reduction of soot and NO_x with a lower amount of exhaust gas recirculation (EGR) as compared to diesel, with NO_x and unburned hydrocarbons (UHCs) generally below Euro 5 and 6 emissions standards [3]. Splitter, using a dual-fuel RCCI approach with gasoline and ethanol, found that UHC emissions were a function of piston crevice and squish geometry, and CO emissions were sensitive to local equivalence ratio [4]. He predicted that by maintaining lean conditions with sufficient fully premixed fuel, simultaneous reductions in heat transfer and incomplete combustion could be achieved.

Butanol has been studied as a promising biofuel to use for PPC due to prolonged ignition delay, low cetane number (CN), and high volatility fuel characteristics. Cheng et al. achieved PPC with butanol-diesel blends, showing soot emissions could be reduced by 70%, while NO_x increased at low loads [5]. Leermakers et al. showed extremely high soot reduction with moderate butanol-diesel blends compared to diesel-only operation using a PPC approach [6]. They showed butanol-diesel blends to be a viable fuel, with stable operation at 50–70% butanol by volume, and 50% average gross indicated efficiency over the whole load range.

Noehre et al. mapped out the operating range for PPC using standard diesel fuel under a range of engine operating conditions including compression ratio (CR), EGR, speed, and boost at loads up to 15 bar gross indicated mean effective pressure (gIMEP) [7]. The combination of low compression ratio, high EGR and engine operation close to stoichiometric conditions resulted in premixed and diffusion combustion that enabled simultaneous NO_x and soot reduction, though a decrease in efficiency and a rise in UHC and CO emissions were noted. They concluded that soot emission reduction was a result of good premixing, long ignition delay, and low local combustion temperature due to high EGR. Additionally, Dempsey et al. found, for a PPC gasoline approach, that when the premixed fuel percentage was low, the NO_x emissions began to increase due to the existence of high equivalence ratio regions in the combustion chamber at the start of combustion (SOC) [8]. Kalghatgi et al. demonstrated the ability to use the soot reduction of PPC for high load extension [9]. Using PPC with gasoline, they reached mean IMEP of 15.95 bar and smoke opacity of 0.33%, while to attain the same level of smoke with diesel fuel, IMEP had to be below 6.5 bar.

Zha et al. compared a high CN jet propulsion “8” (JP-8) fuel with that of a low CN Sasol JP-8 under PPC mode [10]. They found that low CN Sasol JP-8 exhibits longer ignition delay, lower premixed peak heat release, and later combustion phasing compared to JP-8 combustion. Low CN Sasol JP-8 was shown to be a satisfactory under light-load condition, but with higher UHC emissions and lower fuel efficiency compared to high cetane JP-8 fuel. This investigation seeks to expand on the work of Zha et al. in using jet fuel for a PPC approach.

From a military standpoint, due to the significant logistical burden of a dual-fuel RCCI approach, this study focuses on a single-fuel PPC approach. This is consistent with Department of Defense Directive 4140.43, the “One Fuel Forward” initiative, meant to reduce the logistics burden associated with bringing multiple fuels to the battlefield. The fuel used in this study was Navy jet propulsion “5” (JP-5), a fuel commonly used in many Navy and Marine Corps platforms due to its low volatility for safety purposes. Thus, the objective of this work is to work is to principally focus on approaches to quantify the PPC based power improvements based on the high load extension of the exhaust sooting limit in a compression ignition diesel engine as compared to conventional diesel operation.

The engine used in this study was the Waukesha Cooperative Fuels Research (CFR) diesel engine. This indirect injection (IDI) engine has mechanically variable CR, injection amount, and injection timing. For this PPC alternative combustion mode study, the IDI prechamber fuel injection system was supplemented with a Cole-Parmer peristaltic pump delivering jet fuel directly to the intake manifold. In-cylinder pressure was measured with a Kistler 6125C piezoelectric sensor. Engine position was measured using a BEI dual channel shaft encoder coupled directly to the crankshaft. The CFR cetane rating engine operates at a fixed 900 rpm. Experimental details can be found in other Refs. [11,12]. Exhaust emissions (via a heated Unique Heated Products filter) were measured with an Infrared Industries FGA4000S analyzer. Opacity was measured with a Wager 6500 opacity meter. During some of the experiments, exhaust particulate measurements were acquired with a TSI 3090 EEPS unit via a Matter Engineering diluter.

The general categories of components that are present in JP-5 were determined using gas chromatograph (GC)/mass spectrometry analysis. An Agilent 5975 inert mass selective detector was used in conjunction with an Agilent 6890N Gas Chromatograph System under the same conditions utilized in previous studies [13]. The GC was equipped with a Zebron phase ZB-5 ms® column (30 m, 0.25 mm, 0.25 μm, 5% diphenyl-arylene/95% dimethyl polysiloxane) and operated at a helium flowrate of 1.5 mL/min. The compounds were separated using a temperature ramping program that started at 60.0 °C and climbed at a rate of 10 °C/min to 250 °C. An electron impact ionization method was used for the mass spectrometer with an m/z scan from 30 to 600. To prevent the detector from being saturated, all fuel samples were diluted in n-hexane (1/100 dilution), and the mass spectral analysis of the fuel sample was delayed until the solvent peak had passed. The fragmentation patterns of peaks found at several retention times were compared to those found in the National Institute of Standards and Technology (NIST) database to determine potential compounds present.

The speed of sound of the fuels was measured using an Anton Paar DSA 5000 Density and Sound Analyzer. Certified reference standards were used to test the accuracy of these instruments, which were recalibrated if they failed testing. The bulk modulus at 20 °C was calculated by multiplying the density by the speed of sound squared. The other properties were determined using American Society of Testing and Materials techniques specified in the military specifications.

Navy jet propulsion 5 fuel was used in this investigation. This jet fuel has a moderately higher flash point (less volatile) as compared to conventional commercial jet fuels. A summary of key properties is shown in Table 1.

The chromatogram of the fuel is shown in Fig. 1. Based on a NIST database search and comparing to original standards, the major peaks at 5.5, 7.1, 8.7, 10.1, 11.5, and 12.7 min are decane, undecane, dodecane, tridecane, tetradecane, and pentadecane, respectively. Other components identified by the NIST database include branched alkanes, aromatic compounds, alkylcyclohexanes, and indenes. The specific compounds were not confirmed by original standards.

After the CFR engine was warmed up using conventional diesel combustion with Navy jet fuel, the desired CR was adjusted followed by baseline Conventional IDI (Conv. IDI) data collection. Engine load was set starting in the midload range of the diesel CFR (∼5–6 bar gIMEP), then incrementally increased until relatively high levels of CO were observed. At each operating point, a real-time fast heat release analysis was performed in order to maintain the crank angle location for 50% heat release (CAD50) nominally around 10 deg after top center (ATC) with slight start of injection (SOI) adjustments.

After the Conventional IDI fuel–load sweep was performed, the SOI and fuel injection amount (e.g., load) were reset to the condition where opacity was nominally 3%. At this point, two different PPC port jet fuel injection levels were introduced. Data were collected at each operating point. Since JP-5 is very reactive, only a minority of fraction of this jet fuel could be applied to the intake port due to premature combustion during the compression stroke. The maximum port fuel injection level was empirically determined by the loss of SOI control. Another paper by the authors outlines this auto-ignition nature of JP-5 at various compression ratios [14]. It should be noted that this approach does not seek to optimize the PPC combustion by also adjusting SOI, but rather to see the power increase effects with PPC added to a conventional diesel engine without adjustment (e.g., retro-fit).

At each stabilized operating point, 25 s of in-cylinder pressure data were collected at a 50 kHz sampling frequency using MATLAB data acquisition software and National Instruments hardware. A conventional single-zone heat release analysis approach was then applied to the data as outlined in Refs. [15] and [16].

A peristaltic pump was used to deliver port fueling. The pump was calibrated to determine fuel volume as a function of time at various pump settings. The fuel was continuously delivered to the intake pipe runner approximately 25 cm upstream of the cylinder head. In applying the PPC approach, the amount of port fuel injection was steadily increased until the SOC was unable to be controlled by the in-cylinder injection event at SOI due to pre-ignition of the port injected charge, where ignition delay (IGD) was slightly negative (in terms of degrees ATC). Figure 2 shows the two PPC operating fueling amounts used in this study as a function of CR. The upper line (“o” symbols) is the maximum port fuel fraction added at the threshold of port injected fuel pre-ignition. The lower line (diamond symbols) is an operating condition slightly less of maximum where combustion phasing is still controllable with SOI adjustments.

The amount of port fuel injection to the intake manifold was determined from the peristaltic pump calibration. This value was also similar to the air–fuel equivalence ratio (λ) measured from the ETAS-Bosch oxygen sensor at the maximum Conventional IDI operating point (3% opacity) of the same CR to determine the port fuel fraction (port fuel on versus off). As seen in Fig. 2, more fuel was able to be port injected at higher compression ratios since the overall high load Conv. IDI limit became leaner as CR increased (results to be discussed shortly).

Next, Figs. 3 and 4 focus just on the specific CR 16.5 operating point; operation through the range of compression ratios will be discussed later.

Figure 3 compares PPC to Conventional IDI operation at CR16.5. The open blue square symbols with both solid and dashed lines show Conv. IDI operating mode. The solid line connects the data below the opacity threshold; the dashed line data show conventional engine operation in the high load excessively sooty regime. The blue square solid filled symbols shows PPC operation. On the x-axis, gIMEP increases directly with the amount of fuel injected for fixed engine speed. A dashed horizontal red line marks the 3% exhaust opacity threshold defined to be the point of maximum engine load such that engine soot is too great for further enriched operation. A vertical dashed red line marks the engine load (gIMEP) where conventional operation reaches this 3% threshold, approximately 8.5 bar. As seen in Fig. 3 (bottom graph), PPC operation does not reach this 3% threshold until a higher engine load, 9.3 bar, because more fuel was able to be added via port injection without a significant increase in exhaust soot.

The top of Fig. 3 shows that the power for Conv. IDI operation at the opacity threshold (lightly dashed red line) was 3.8 kW, while the maximum PPC point achieved 4.25 kW, an 11% improvement. Exhaust CO (Fig. 3 second from bottom) had similarly low levels between the operating modes at the opacity threshold, indicating nearly complete combustion (as CO levels are just beginning a dramatic rise with Conv. IDI).

Exhaust oxygen levels (Fig. 3 middle) were lower in the PPC case compared to the Conv. IDI case at the opacity limit indicating that more of the available oxygen was being consumed in partially premixed combustion due to the increased fuel addition and port injection air–fuel mixing. Thermal efficiency (Fig. 3 second from top) was moderately lower in the PPC case compared to the Conv. IDI case at the opacity limit due to the increase in specific fuel consumption. However, as seen in the Conv. IDI operating points above the opacity threshold, there was a similar decrease in thermal efficiency. Thus, this loss in efficiency is not a problem endemic only to PPC operation.

Figure 4 continues the comparison of PPC to Conventional IDI operation at CR16.5. Ignition delay (Fig. 4 bottom) was insensitive for Conv. IDI operation across the range of engine loads. As discussed in Fig. 2, PPC was set such that IGD was near zero or negative in degrees ATC, limited by pre-ignition of the fuel primed mixture from port injection.

The 10% heat release location (CAD10) shown second from the bottom in Fig. 4 was defined as the SOC in this study. For Conv. IDI operation, CAD10 steadily advanced as gIMEP increased. This is because SOI was advanced to inject more fuel to achieve higher gIMEP keeping CAD50 approximately consistent. For PPC, SOI was set to −10 deg ATC (essentially stock Conv. IDI setting—not optimized for PPC), and CAD10 (and thus SOC) was reached almost immediately after injection due to pre-mixing and heat release of the port injected fuel. The 50% heat release location (CAD50, Fig. 4 middle) was kept near 8–10 deg ATC for both operating modes across the range of engine loads to retain ideal combustion phasing for best torque. As gIMEP increased, especially with PPC, CAD50 advanced due to the increased fuel injection.

Exhaust carbon dioxide (CO₂, Fig. 4 second from top) increased with gIMEP because as more fuel was added to achieve higher engine loads, more CO₂ was formed in combustion. NO_x, shown at the top of Fig. 4, followed the characteristic curve for Conv. IDI operation. NO_x peaked at midload due to the steady increase in combustion temperature and steady decrease in available oxygen as gIMEP increases. NO_x for the PPC case was slightly elevated due to a modest increase in combustion temperature, shown later from slightly greater peak combustion pressures.

A characteristic in-cylinder pressure and cumulative heat release curve for both Conv. IDI (9.2 bar) and PPC (9.3 bar) are shown next in Fig. 5. The PPC data (green, also labeled) clearly show an early energy release period (∼330 deg) followed by the main fuel injection energy release period taking off at ∼345 deg. Due to the minority fuel-energy fraction of the PPC port fueling, only a modest low level of early PPC energy release is observed. The Conv. IDI heat release shows some slight evaporative cooling (∼350 deg) followed by the fuel injection-driven energy release period.

Figures 6–18 compare PPC and Conventional IDI engine operation over four compression ratios. As above, solid symbols indicate PPC operation, open symbols indicate Conventional IDI operation. The start of the dashed line indicates the point at which Conv. IDI operation reaches the 3% exhaust opacity threshold similar to the above composite figures at CR 16.5. In the following, circle symbols denote CR14, square symbols, CR16.5, diamond symbols, CR18.5, and triangle symbols are CR21.5.

The engine soot limit was set at 3% exhaust opacity. Any further increase in engine load, shown by the dashed lines in Fig. 6, was considered outside of the normal Conv. IDI operating range of the engine. Data were collected at these operating points for the sake of comparison to high load PPC operation. Figure 6 shows that significant sooting was delayed until much higher loads using PPC. PPC allowed for high load gIMEP extension past the maximum point of Conv. IDI engine operation. Exhaust opacity for the Conventional IDI operating points at high gIMEP were significantly greater than the PPC operating points at the same load, across the range of compression ratios.

Figure 7 shows an advance in SOI for all cases with increasing gIMEP such that CAD50 was kept nominally constant ATC for optimal torque-combustion phasing. SOI was advanced for higher CRs, because at higher CRs, IGD was shorter due to the elevated in-cylinder pressures. Shorter IGD led to less time for premixing and thus longer 10–90% burn durations, shown in Fig. 10. While PPC operating points maintained relatively similar SOI to Conv. IDI operation, a −10 deg ATC SOI is likely the earliest possible timing due to the onset of pre-ignition.

Shown in Fig. 8, CAD10 is roughly 6 degrees earlier for PPC. Figure 8 shows CAD10 following the same trends as SOI. For higher gIMEP, fuel injection amount was increased; thus, SOI must advance leading to earlier CAD10 and SOC. At higher CRs, CAD10 advanced due to increased pressures at SOI leading to earlier SOC.

Ignition delay, in Fig. 9, was defined as the difference between SOI and CAD10 (SOC). For Conv. IDI operation, CR14 was expected to have the coolest in-cylinder pressures/temperatures during compression and thus the longest IGD. The PPC operating points at CR16.5, 18.5, 21.5 had IGDs of slightly negative or near zero; thus the premixed fuel began to burn before or during the direct injection event. PPC at CR14 had IGDs close to those of Conv. IDI operation.

The burn duration, shown in Fig. 10, increased with increasing CR and increasing gIMEP. To achieve higher engine loads, more fuel was injected leading to a longer combustion event. At higher CRs, IGD was shorter, and these shorter IGDs led to less time for premixing and thus longer burn durations. Despite the port injection, PPC operating points had even longer burn durations. The authors hypothesize that the port injected charge was burning early (indicated by the negative IGD), thereby consuming oxygen in the direct injection flame region. This small oxygen deficiency led to a slower, diffusion-dominated flame.

Figure 11 shows increasing peak in-cylinder pressures as CR and load (fueling) were increased, in accordance with that expected by the ideal gas law. Peak combustion temperatures normally tend to increase with peak combustion pressures for a given CR. PPC peak pressures were close to those of Conv. IDI operation. Thus, it would be expected that NO_x increases with PPC due to the higher peak pressures. However, as seen in Fig. 12, this was not the case.

Figure 12 shows that exhaust NO_x concentration followed the characteristic curve for Conv. IDI operation as a function of load. NO_x peaked at mid-load due to the steady increase in combustion temperature and steady decrease in available oxygen as gIMEP increased. PPC, operating at high loads, followed this same NO_x trend as Conv. IDI operating above the 3% opacity threshold (dashed lines). For PPC, the authors believe the premixed charge consumes oxygen in combustion outside of the flame zone, in the cylinder periphery, indicated by the short IGDs, leading to the low levels of NO_x. Overall, as CR increases, exhaust NO_x concentrations decrease, despite the greater peak pressures at high CR. The authors hypothesize that (to be shown in Fig. 14) at higher CRs, in-cylinder temperatures do not increase, but rather moderately decrease, with the increase in CR and peak pressures. This is because, for this fixed displacement engine, the higher compression ratios lead to smaller combustion chamber volumes, which tend to lower combustion temperatures (to be further discussed shortly). Thus, the moderately lower temperature combustion event at increased CR reduced NO_x levels.

The overall equivalence ratio increased at higher gIMEP due to the greater fuel addition. Because PPC allowed for the addition of more fuel during port injection, the overall equivalence ratio was even higher for the PPC cases, shown in Fig. 13. At CR14, PPC allowed operation near stoichiometric conditions. However, shown in Fig. 16, at CR14, thermal efficiency was significantly lower due to the decrease in expansion work during the combustion stroke.

The equivalence ratio at the soot limit (nominally 3%) is shown in Fig. 14 as a function of CR. It is seen that as CR increases, the corresponding maximum fueling (e.g., equivalence ratio) that can be fueled must decrease significantly. When the peak combustion pressure is converted to an average charge temperature using the ideal gas law and cylinder volume at the time of combustion, the peak charge temperature is predicted (air only gas constant), shown in Fig. 14. Because the combustion chamber volume decreases with an increasing CR for a fixed displaced volume, the peak temperature also decreases as the air flow rate was constant with increasing CR. The increase in combustion pressure with increasing CR does not offset the reduced combustion chamber volume; thus, average combustion temperatures moderately fall with increasing CR. This helps explain why the NO_x levels decreased with increasing CR.

As this result is somewhat counterintuitive, more detail is shown in Fig. 15: CR21.5 (solid line) and CR14 (dashed line) in-cylinder pressure, volume, and average predicted temperature A representative in-cylinder pressure trace is shown for from the 7.15 bar gIMEP cases in the top segment. For reference, a dotted ideal polytropic (n = 1.33) reference line is shown for both compression ratios. As expected, the CR21.5 case has much higher compression and combustion pressure as compared to the CR14 case. The middle segment of this figure shows the total cylinder volume as a function of engine position. The CR21.5 case has a smaller volume throughout the cycle due to the smaller clearance volume at this high CR. The lower segment shows a “single zone” temperature estimate using the ideal gas law. It is seen that during compression, both cases have similar predicted charge temperatures. Near the later phase of compression, the CR21.5 case is hotter than the CR14 case, with combustion starting slightly earlier as described above. Then, quite interestingly, the CR14 single zone—average charge temperature during combustion exceeds that of the CR21.5 case. This result is due to the larger charge volume (nearly 1.6 times larger than the high CR case) that exists at CR14 despite its lower overall pressure.

Next, on looking at engine indicated efficiency, the PPC cases resulted in a moderate decrease in thermal efficiency at high loads compared to the Conv. IDI high load limit, shown in Fig. 16. This is because at higher loads, there were increased heat losses as more fuel was added and there was a lower, less advantageous ratio of specific heats at these PPC richer conditions. However, as seen in the Conventional IDI operating points above the opacity threshold, there was a similar decrease in thermal efficiency. Thus, this loss in efficiency is not a problem endemic only to PPC operation. It is interesting to note that the higher compression ratios have very similar thermal efficiencies. While higher compression ratios allow for increased expansion work, it also results in higher compression/combustion temperatures, which increase heat losses. As CR increases, these effects offset. The effects of engine efficiency with engine-out NO_x are shown in Fig. 17. In general, NO_x peaks near the highest measured efficiency (for a given CR). This corresponds to a mid-high load point. NO_x and efficiency then fall with increased enrichment and with PPC. The increased enrichment reduces available oxygen necessary for NO_x formation.

Exhaust carbon monoxide concentration in Fig. 18 follows similar trends as exhaust opacity in Fig. 6. The port injection event of PPC allowed for greater air–fuel mixing prior to SOC, leading to more complete combustion and a reduction of CO at high loads. Time for mixing during Conventional IDI was limited due to engine speed. Port injection allowed for mixing in the intake manifold and during the intake and compression strokes, leading to a more partially “homogenous” premixed charge prior to the main injection event.

Figure 19 shows the exhaust soot-particulate concentration (numbers of particles per volume) as a function of particle size. Due to instrument saturation limits, these data were collected at high load just near or below the minimum detection limit of the opacity meter. The integrated results for each curve are converted to a total mass concentration assuming spherical carbon particles. Two cases are run for both Conv. IDI as well as PPC at a CR14, with equivalence ratios of 0.85 and 0.89. Only the 0.89 case with Conv. IDI has just exceeded the minimum opacity threshold (∼1%). The two PPC cases were expected to be well below the minimum opacity detection level.

Figure 20 illustrates the key finding of this study. More fuel was able to be added using PPC at higher compression ratios because the 3% exhaust opacity soot limit was delayed until higher engine loads compared to Conventional IDI operation. At fixed engine speed, this torque improvement translated into a moderate power improvement over Conventional IDI diffusion operation. Power increased from 3% to 23% at the highest CR.

This study applied a PPC approach with a minority fraction port fuel injection in order to extend the high load limit based on exhaust soot opacity. Jet fuel was used as the same fuel for both port and in-cylinder injection. It was seen that the Conventional IDI soot limit occurred at leaner overall maximum fuel–air equivalence ratios as CR was increased. This afforded a greater opportunity to increase port injected fuel levels with increased CR. It was thus seen that power levels were then able to increase from 5% to 23% (at CR 21.5) over conventional diesel combustion by applying port injected companion fuel (PPC) without a soot opacity penalty.

This research was funded by the Office of Naval Research with Dr. Maria Medeiros as Program Manager. The authors would also like to thank Mr. Steve Galindo for his assistance in the experimental setup.

• Office of Naval Research Global (N0001417WX00892).

ATC =

after top center

CAD10 =

crank angle for 10% heat release

CAD50 =

crank angle for 50% heat release

CFR =

Cooperative Fuels Research

Conv. IDI =

conventional indirect injection

CN =

cetane number

CO =

carbon monoxide

CO₂ =

carbon dioxide

CR =

compression ratio

EGR =

exhaust gas recirculation

GC =

gas chromatograph

gIMEP =

gross indicated mean effective pressure

IDI =

indirect injection

IGD =

ignition delay

IMEP =

indicated mean effective pressure

JP-5 =

Navy jet propulsion “5”

JP-8 =

Navy jet propulsion “8”

NIST =

National Institute of Standards and Technology

NO_x =

oxides of nitrogen

PPC =

partially premixed combustion

RCCI =

reactivity controlled compression ignition

rpm =

revolutions per minute

SOC =

start of combustion

SOI =

start of injection

UHC =

unburned hydrocarbon