1 Introduction

Regulations on exhaust emissions and fuel consumption in the automotive industry to mitigate global warming have long been discussed and continue to be strengthened. The European Commission has strengthened the post-2020 CO2 standards for new passenger cars and vans, known as Fit for 55. This regulation requires a reduction in CO2 emissions by up to 55% by 2030 compared to the 2021 baseline (European Commission, 2020). Therefore, for spark-ignited (SI) engines, research has continued to increase thermal efficiency by increasing the compression ratio to cope with enhanced regulations. However, the increase in temperature and pressure of unburned gas due to the high compression ratio causes knocking phenomena in the engine (Wang et al., 2017). To prevent knocking, which can cause severe damage to the engine cylinder and piston, the spark timing needs to be retarded, resulting in the inability to achieve maximum brake torque (MBT) timing. This results in heat loss, which acts as a major obstacle to increasing thermal efficiency.

Reducing the cylinder gas temperature, increasing the flame propagation speed, or changing the fuel characteristics can prevent knocking phenomena (Cho et al., 2018). Among them, water injection (WI) is a technique that reduces knocking by reducing the temperature inside the cylinder. The heat of vaporization of water is more than six times that of gasoline, which is able to significantly reduce the temperature of the unburned gas inside the cylinder (Kim et al., 2016). In addition, the specific heat of water, which is approximately twice that of fresh air, has an additional dilution effect after evaporation of water is completed (Hoppe et al., 2016; Worm, 2017). By using these characteristics of water, WI can reduce knocking, advance the ignition timing, and address the fuel enrichment problem under high load and RPM conditions by reducing the in-cylinder gas temperature. Through this procedure, as well as applying high compression ratios and increasing the maximum torque, improved fuel consumption rates can be achieved.

WI faded away after its application in aircraft engines in the 1940s (Daggett et al., 2004). However, with the development of direct fuel injectors and stricter CO2 regulations, research on the effect of WI in automobile engines has regained momentum, as demonstrated by BMW’s application of WI technology to the M4 GTS in 2016 (Durst et al., 2017). Pauer et al. (2016) studied the potential of applying port water injection (PWI) in turbocharged gasoline direct-injection (T-GDI) engines, which reduced fuel enrichment and knocking under low-load operation conditions, thereby improving thermal efficiency by up to 13%. Although injecting water into the intake port can take advantage of lowered intake temperature and improved volumetric efficiency, there is an issue with incomplete evaporation of water and wall wetting in the intake port, limiting the amount of water that can be injected (Worm, 2017). In contrast, direct water injection (DWI) allows for maximum evaporation during compression and overcomes the limitation on WI amount (Cordier et al., 2019; Singh et al., 2020). Hoppe et al. (2016) determined an efficiency improvement of 3.8% in the sweet spot region and up to 16% at full load by applying a dual direct-injection system of water and gasoline into the combustion chamber. Cordier et al. (2019) compared PWI and DWI in a T-GDI engine and observed that DWI is more effective in terms of combustion phase advancement and exhaust temperature reduction at the same WI amount.

Previous research on engines with WI have primarily conducted experiments on turbocharged engine configurations that can maximize water evaporation at high intake pressures (Wan et al., 2021). On the other hand, Kim et al. (2016) studied the effects of DWI in a naturally aspirated (NA) engine with a fixed WI timing at 120° crank angle (CA) before the top dead center (TDC). WI has a load expansion up to 14% and a BSFC reduction of 16–17% at full-load conditions, and the effect of injected water mass was investigated. Thus, it can be sufficiently effective even for NA engine under the high-load or full-load conditions. In addition, replacement or combination of EGR and Lean burn with the same diluent gas effect could be possible (Hoppe et al., 2016). However, there is a lack of analysis of the effect of water evaporation induced by WI parameters on NA engine configurations, where evaporation and atomization may not occur as effectively. Research on water evaporation has primarily been conducted through CFD simulations. Bhagat et al. (2013) investigated the evaporation and wall film characteristics by DWI timing using water spray visualization experiments and CFD simulations in a constant combustion chamber. Battistoni et al. (2017) simulated the effect of WI based on the position of the PWI injector, proposing a quasi-direct WI injector near the intake valve. Vacca et al. (2019) investigated the influence of PWI and DWI on water evaporation, wall wetting, and liquid water residual in a 3D CFD virtual test bench. Raut and Mallikarjuna (2020) confirmed the water evaporation rate by the water fuel ratio (WFR) and the effect on combustion and exhaust emissions according to the nozzle configuration of the DWI injector. Wang et al. (2022) also studied the effects of the position of the DWI injector.

The goal of this study is to focus on WI in naturally aspirated SI engines and to investigate the effects of WI variables on fuel efficiency by applying PFI and DWI configurations. First, spray visualization was conducted to understand the spray characteristics during high-pressure WI using a commercial GDI injector. Engine experiments were carried out under the same torque conditions to analyze the effects. The efficiency improvement was measured under knock limit spark advance (KLSA) conditions by adjusting the WI timing and amount. The compatibility of replacing or combining EGR, which has the same dilution gas effect, was also analyzed. Finally, WI application to lean combustion was investigated in terms of fuel efficiency and combustion stability.

2 Experimental Setup

2.1 Spray Visualization

In the spray visualization experiment, the Mie scattering technique, which enables liquid-phase measurements, was applied. High-speed cameras were used to capture images at a rate of 50,000 frames per second (FPS). The measurement range was 114.8 × 114.8 mm, taking into account the shape of the engine combustion chamber. The test was performed in ambient conditions simply in air, and the experimental schematic of spray visualization is shown in Fig. 1. A commercial GDI injector was used with injection pressures ranging from 100 to 150 bar. The injection duration was adjusted to 0.4–2 ms, while the injector tip temperature was compared at 25 °C and 90 °C to simulate the engine combustion chamber conditions. Gasoline spray measurements were also conducted under the same conditions to investigate any differences. The detailed experimental conditions are summarized in Tables 1 and 2.

Table 1 Experimental setup of the spray visualization
Full size table
Table 2 Experimental conditions of the spray visualization experiment
Full size table
Fig. 1
figure 1

Schematic of the spray visualization experiments (in case of gasoline injection)

Full size image

2.2 Engine Experiments

The engine experiments were conducted on a 4-cylinder 2.0 L naturally aspirated GDI engine converted to a PFI engine with EGR. The GDI injection system was connected to a water supply system for DWI. An Atkinson cycle camshaft was applied with a compression ratio of 13.5. The details of the engine specifications are listed in Table 3.

Table 3 Engine specification
Full size table

The engine experiment was configured as shown in Fig. 2 to determine engine performance and collected ECU data and various sensor data. An AVL ELIN dynamometer system (190 kW) was used for the engine experiments. ECU control parameters were controlled through ETAS INCA software using ETAS ES-591 equipment. In-cylinder pressure was measured through a Kistler Type 6115CF spark plug integrated pressure sensor. Combustion data were monitored in real time using a commercial Kistler Kibox combustion analyzer during the experiments. The experimental data were obtained by 200-cycle averaged data.

Fig. 2
figure 2

Schematic of the engine experiments

Full size image

WI was controlled through National Instruments (NI) LabVIEW software using an NI PXI 8119 DAQ controller and injector drivers. Water was supplied using a filter system including a reverse osmosis filter. Haskel’s MS-36 pneumatic pump was used to pressurize, and a relief valve was used to maintain a constant pressure. In addition, a reservoir was installed in the high-pressure water supply line to reduce pressure oscillations caused by the pump. The final high-pressure water was injected through a commercial GDI common rail and injectors.

The injection parameters were investigated under engine operation conditions of 1600 RPM, BMEP of 7 bar and 2000 RPM, and BMEP of 7 bar for lean combustion. The WI timing was varied by 50° CA based on the start of injection (SOI), and the injection amount was adjusted according to the water-to-fuel (W/F) ratio from 0.25 to 1. The injection pressure was fixed at 100 bar (Table 4).

Table 4 Engine experimental conditions
Full size table

3 Results and Discussion

3.1 Spray Visualization of Water Injection

Figure 3 shows a comparison of the spray shapes of gasoline and water from a commercial GDI injector. Spray plumes with six holes are maintained and considered to be normally sprayed, similar to gasoline. When the temperature increases, the water droplets, which have a clearly visible shape at 25 °C, look blurry due to active atomization. In the image of gasoline injection at 90 °C, the penetration length is shortened and the spray shape collapses as one plume due to flash boiling. Flash boiling is a phenomenon in which droplets are atomized due to rapid boiling when the overheated liquid is exposed to a pressure lower than the saturated vapor pressure (Zeng et al., 2012).

Fig. 3
figure 3

Mie scattering image of spray development injected from a 6-hole commercial GDI injector. Injection pressure of 100 bar, injection duration of 1 ms (left: gasoline, right: water)

Full size image

The raw images were postprocessed using MATLAB software for quantitative analysis, specifically for the estimation of the spray penetration length and liquid spray area. A longer penetration length can lead to issues such as wall wetting and uneven distribution due to incomplete fuel evaporation (Schulz & Beyrau, 2018). According to Fig. 4, the penetration length did not show any significant change within the measured range of 106.94 mm with respect to the injection duration. Although the liquid spray area tended to increase with the injection duration, no significant difference was observed between injection durations of 1.0 ms and 2.0 ms. Therefore, the injected water was not completely evaporated, leaving a liquid volume regardless of the injection duration. According to Fig. 5, both the penetration length and the spray area tended to increase with increasing injection pressure due to the higher momentum. (Mitroglou et al., 2006) The liquid spray area decreased significantly with increasing injector tip temperature, indicating that the cylinder wall temperature should be sufficiently warmed up for water evaporation.

Fig. 4
figure 4

Penetration length and spray area of water injection. Injection pressure of 100 bar, injector tip temperature of 25 °C

Full size image
Fig. 5
figure 5

Penetration length and spray area of water injection. Injection duration of 2 ms

Full size image

Comparing the quantitative values of gasoline and water in Fig. 6, water has poorer evaporation than gasoline. This result could be attributed to the thermodynamic properties of water and gasoline. (Vacca et al., 2019) The thermodynamic properties at 25 °C are shown in Table 5 (Wang et al., 2006). First as shown in Fig. 7 (Badawy et al., 2022; Moran et al., 2017), under conditions below 100 °C, the saturation pressure of water is lower than that of gasoline, which reduces the evaporation of water under the experimental conditions with low ambient air temperature. In addition, the surface tension of water is three times higher than that of gasoline, which hinders the atomization of water droplets, as shown in Fig. 3. In addition, a higher specific heat of water delays reaching the conditions needed for evaporation. All these properties can cause water evaporation even under warm-up (90 °C) temperatures to occur slower than that of gasoline under cold-start conditions (25 °C). Based on the above results, under standard temperature and atmospheric pressure conditions, it was confirmed that water droplets remained after the end of injection due to incomplete evaporation regardless of the injection conditions. When applied under engine conditions, the measured range of 2.5 ms corresponds to a duration of 24° CA at 1600 RPM. Through this, wall wetting occurs even with a short injection time when WI is performed in an engine cylinder. This indicates that optimization of injection timing and fuel temperature is an important factor.

Fig. 6
figure 6

Penetration length and spray area. Injection pressure of 100 bar, injection duration of 0.5 ms

Full size image
Table 5 Thermodynamic properties of isooctane, gasoline, and water
Full size table
Fig. 7
figure 7

Saturated pressure of isooctane, gasoline, and water

Full size image

3.2 Effect of Water Injection Timing and Duration

The effect of the WI timing based on the start of injection (SOI) was investigated by analysis of spark timing, mass fraction burned (MFB) 50, and BSFC reduction compared to the base condition without WI. The spark timing of the base condition WI is 3.4° CA bTDC, MFB 50 is 29.0° CA aTDC, and the coefficient of variation (CoV) is 2.81%.

Figure 8 shows that the spark timing advance in WI during the intake and compression processes was approximately 2–3° CA, while MFB 50 tended to be retarded compared to the base condition due to the dilution gas effect. Accordingly, BSFC reduction was also insignificant, approximately 1%. However, WI during the exhaust process after TDC showed a spark timing advance of up to 5.6° CA and an MFB 50 advance of 6.7° CA. The BSFC was reduced up to 4.95% under SOI at 300° CA aTDC, the rear end of the exhaust stroke. Figure 9 shows the results of WI by increasing the WFR to 0.47 (approximately 7.7 CA of injection duration). An increase in the WI amount enhanced the injection timing tendency, resulting in a 6.1% BSFC reduction under the same SOI conditions. In previous studies, the optimal WI timing was estimated to be 150–100° CA bTDC at the compression stroke after the intake valve closed (Cordier et al., 2019; Hoppe et al., 2016). This study identified many differences in the optimal WI timing, and the possible reasons are as follows.

Fig. 8
figure 8

Water injection timing effect on combustion characteristics and fuel consumption. Engine speed of 1600 RPM, BMEP of 7 bar, injection pressure of 100 bar, of WFR 0.25

Full size image
Fig. 9
figure 9

Water injection timing effect on combustion characteristics and fuel consumption. Engine speed of 1600 RPM, BMEP of 7 bar, injection pressure of 100 bar, WFR of 0.47

Full size image

During the intake and compression strokes, low cylinder pressure and temperature can deteriorate water evaporation (Kim et al., 2016). For detailed analysis, the averaged cylinder pressure curve was extracted from a computational fluid dynamics (CFD) simulation using the commercial software CONVERGE under the same conditions as the experiments. The κ-ε RNG model for turbulence estimation and the SAGE model for combustion were utilized for the simulation. In Fig. 10, the averaged cylinder pressure curve from 286° CA bTDC to 102° CA bTDC is located on the left side of the water saturation pressure curve, which is the condensation zone. This region is similar to the environmental conditions in the spray experiment, where water evaporation does not occur for at least 24° CA. Therefore, immediate evaporation of water may not take place under WI at the condensation zone, and water droplets would remain until TDC, which is the start of combustion (SOC). Bhagat’s CFD results showed that 38% of injected water remains in liquid form near TDC when water is injected at 90° CA bTDC (Bhagat et al., 2013).

Fig. 10
figure 10

Temperature–pressure curve of water saturation and in-cylinder gas mixture. Engine speed of 1600 RPM, BMEP of 7 bar

Full size image

Furthermore, Vacca et al. (2019) explained that the deterioration of water evaporation is caused by the high heat of vaporization of water. A high heat of vaporization significantly reduces the temperature of the surrounding air–fuel mixture, decreasing the saturated vapor pressure and preventing further evaporation of liquid droplets. Vacca’s simulation results showed that only 16% of water evaporated until the ignition time when water was injected into the intake port at 390° CA bTDC with WFR 0.3. With the same thermodynamic properties, applying the heat of vaporization of gasoline increases the evaporation rate to 40%, proving that the impact of the heat of vaporization is quite large.

Last, due to the direct injector’s position being on the side, wall wetting is more likely to occur around the spark plug or on the piston. As a result, the flame may not be properly propagated or combustion may be unstable due to wall wetting or locally distributed liquid water droplets (Raut & Mallikarjuna, 2018). The high CoV observed for WI at the intake and compression stroke shown in Figs. 8 and 9 supports the analysis.

On the other hand, in conditions with WI during the exhaust stroke, water evaporation is expected to occur sufficiently due to the high exhaust temperature. However, since most of the injected water is discharged from the cylinder until the exhaust valve is closed, the effect of reducing the in-cylinder temperature may not be sufficient. Using the CFD simulation, it is confirmed that 3.76 mg of water remains in the cylinder when a total of 11 mg of water is injected at 300° CA aTDC. The proportion of exhausted water of the total amount injected is 65.8%.

The effect of the WI amount was investigated at 300° CA bTDC, which showed the optimal timing effect, as shown in Fig. 11. The increase in the WI amount led to a spark timing advance and an improvement in the fuel efficiency. This could be attributed to the combustion phase not yet reaching the MBT and exhausted water amount during the exhaust stroke, which prevented reaching the maximized effect of water injection.

Fig. 11
figure 11

Water injection amount effect on combustion characteristics and fuel consumption. Engine speed of 1600 RPM, BMEP of 7 bar, injection pressure of 100 bar, SOI 300° CA bTDC

Full size image

3.3 Synergistic Effect of Water Injection with EGR

In the previous section, WI during the exhaust stroke improved the fuel efficiency, but the effect could not be maximized due to discharged water. It is necessary to keep the residual water amount inside the cylinder just before SOC to maximize the cooling effect. There are some solutions for this, such as optimizing the injector configurations and position, increasing the tumble ratio, and enhancing the cylinder gas flow through the use of a charger (Bhagat et al., 2013; Raut & Mallikarjuna, 2020). However, these approaches require modifications of the engine configuration. In this study, the combination with EGR is used as a means to improve water evaporation during the intake and compression stroke without any modifications of engine configuration.

Figure 12 presents the results of the experiment by varying the WI timing under the same engine operation conditions with a WFR of 0.47 and an EGR ratio of 10%. In the EGR-only case, the EGR ratio is 20.24%. Unlike the WI only case, the spark timing is further advanced during the intake and compression strokes. The most advanced spark timing is WI at 250° CA bTDC, and the BSFC showed the most efficiency at 200–150° CA bTDC. This can be attributed to the significant increase in intake air temperature due to EGR, which enhances the evaporation of water during the intake and compression processes compared to that without EGR. The combination of WI and EGR improves fuel efficiency by up to 9.14% compared to the base condition and by 3.05% compared to EGR alone in terms of BSFC reduction.

Fig. 12
figure 12

Water injection timing effect on combustion characteristics and fuel consumption. Engine speed of 1600 RPM, BMEP of 7 bar, injection pressure of 100 bar, WFR of 0.48, EGR ratio of 10%

Full size image

Under WI at 200° CA bTDC, which shows the best fuel efficiency, the effects of the WI amount and EGR rate were investigated. The increased WI amount advances the spark timing, as shown in Fig. 13, but it also increases the burn duration due to the higher dilution rates, resulting in the efficiency reaching a saturation point (Cordier et al., 2019). At a WI amount of WFR 0.47, the effect of the EGR rate does not show any significant fuel efficiency improvement. It can be considered that the dilution effect has already been saturated at this WFR condition. This indicates the need for an appropriate combination of WI amount and EGR ratio, which can be economically achieved through numerical analysis, such as 1D simulation (Bozza et al., 2016). However, as evident from the spray image and engine experimental results of this study, a thorough validation of the evaporation of WI is necessary. Helmich et al. (2021) developed a 0D WI model reflecting fundamental thermodynamic processes for water evaporation under PWI conditions, and if a similar study is conducted under DWI conditions, it can greatly contribute to model-based DWI optimization.

Fig. 13
figure 13

BSFC reduction of water injection with EGR by WI duration (10% of fixed EGR ratio) and EGR rate (0.47 of fixed WFR). Engine speed of 1600 RPM, BMEP of 7 bar, injection pressure of 100 bar

Full size image

3.4 Synergistic Effect of Water Injection with a Lean Burn

The combination with lean combustion, which has another dilution gas effect, was investigated. The target engine can be operated up to λ = 1.4 under the conditions of an engine speed of 2,000 RPM and BMEP of 7 bar. The WI timing was fixed at 300° CA aTDC.

Under the conditions with WI, as shown in Fig. 14, the increase in lambda advanced the spark timing, which led to a BSFC reduction, but the combustion instability indicated by the CoV was increased due to the increased combustion duration. With WI, the fuel consumption at the same lambda condition decreased by an average of 1.8%, and the CoV decreased by an average of 0.7% compared to that without WI. The above results indicated that where the combustion phase did not reach the MBT area, it was possible to improve fuel efficiency and combustion stability. In Fig. 15, the effect of increasing the WI amount is examined. Similar to the results of the integration with EGR and WI, the effects of spark timing advance and increased combustion duration trade off each other, reaching saturation of BSFC reduction. The maximum BSFC reduction as 8.2%, and the optimized WFR as between 0.36 and 0.54 under lambda values of 1.3 to 1.4.

Fig. 14
figure 14

Water injection timing effect on combustion characteristics and fuel consumption under the lean burn condition. Engine speed of 2000 RPM, BMEP of 7 bar, injection pressure of 100 bar, WI SOI 300° CA aTDC, WFR of 0.53

Full size image
Fig. 15
figure 15

Water injection amount effect on fuel consumption under lean burn conditions. Engine speed of 2000 RPM, BMEP of 7 bar, injection pressure of 100 bar, WI SOI 300° CA aTDC

Full size image

4 Conclusion

This paper provides the evaporation characteristics of WI through spray visualization and the effects of WI in terms of combustion characteristics and BSFC in a NA engine. The potential combination of EGR or lean burn was also investigated. The results of this study are summarized as follows.

In the spray visualization, water evaporation was not completed at least 24 CA under the atmospheric pressure conditions of NA engines regardless of the WI amount. Even in warm-up conditions with a water temperature of 90 °C, water evaporated slower than the cold-start temperature (25 °C) of gasoline.

In the NA PFI engine experiment with DWI, the optimal WI timing for fuel efficiency improvement was shown at the late exhaust stroke due to deterioration of water evaporation. The advance of spark timing and combustion phase through knock mitigation by WI were confirmed, but the effect was estimated to be reduced as water was discharged during the exhaust stroke.

Through the integration of EGR, the optimal timing of WI moved to the intake and compression stroke due to improved evaporation of water by increasing the intake air mixture temperature. By adjusting the EGR rate and WI amount, the BSFC reduction was confirmed to be up to 9.3%.

With WI in lean combustion, improving the spark timing advance compared to the increase in combustion duration enabled an additional fuel consumption reduction of a maximum 8.2% decrease in BSFC.