Wang Jun; Yang He; Hu Xiaoming; Song Haiqing; Zhang Ran; Tian Huayu

(SINOPEC Research Institute of Petroleum Processing, Beijing 100083)

Abstract: In this study, the spray auto-ignition properties of binary primary reference fuels (PRFs) of 2,2,4-trimethylpentane and n-heptane with different research octane numbers (RONs) were measured according to the industry standard NB/SH/T 6035 to determine their ignition delay times at various initial temperatures. Furthermore, the auto-ignition properties were investigated after blending the PRFs with various amounts of ethanol. The results revealed a very good correlation between the derived cetane number and the RON for the PRFs in both the presence and absence of ethanol. In addition, a concept of ignition delay sensitivity was developed for ethanol-containing fuels that exhibited a close relationship with the octane sensitivity, which is defined as the RON minus the motor octane number (MON). Finally, the developed method was applied to conveniently estimate the RON and MON values of several ethanol-containing fuels by simply measuring their autoignition properties.

Key words: ethanol, spray auto-ignition, ignition delay, octane number, sensitivity

1 Introduction

Ethanol is the most widely used sustainable and renewable biofuel and is typically blended directly with commercial gasoline as part of the effort to address the increasing energy and environmental issues associated with fossil fuels[1-2]. Owing to its special physical and chemical properties, the combustion of ethanol differs in many ways from that of conventional gasoline. Numerous theoretical and experimental studies have been conducted to investigate the combustion chemistry and reaction kinetics of alcohol-based fuels[3-6]. A key advantage of ethanol blending is that it improves the anti-knock quality,which allows for the operation of spark-ignition gasoline engines at a higher compression ratio[7]. Furthermore, the addition of ethanol can reduce the emission of pollutants such as carbon monoxide and particulate matter[8-10].The higher latent heat of vaporization and lower flame temperature may also be beneficial for decreasing thermal losses and increasing power output[11-12].

With recent advances in engine combustion technologies,several new strategies such as homogenous charge compression ignition and gasoline direct injection compression ignition have been proposed[13-14]. These strategies utilize the concept of low-temperature combustion,and the fuel auto-ignition properties play a crucial role in combustion control. Gasoline surrogates composed of single or multiple components are often used to simulate the behavior of gasoline in a conventional internal combustion engine, among which primary reference fuels (PRFs) based on binary blends of 2,2,4-trimethylpentane andn-heptane have been widely adopted[15-16]. However, it has yet to be established whether these surrogates are suitable for predicting the behavior of gasoline and alternative biofuels in the newly developed combustion modes. Fundamental investigations of the chemical kinetics of fuel combustion are essential for fuel/engine co-optimization.

The ignition properties of a fuel are a key indicator of its combustion chemical kinetics. In compression-ignition diesel engines, auto-ignition is rated by the cetane number(CN). In spark-ignition gasoline engines, engine knock is caused by the auto-ignition of the end-gas mixture.The tendency of a fuel to auto-ignite is described by its research octane number (RON) and motor octane number(MON). Therefore, both the CN and the octane numbers reflect the auto-ignition properties of a fuel, and they usually exhibit a trade-off relationship. Ethanol has a higher octane rating compared with gasoline[17]. When ethanol is blended with gasoline, the octane responses are always non-linear and the interactions are not well understood[18]. Furthermore, the difference between the RON and MON, which is defined as the octane number sensitivity (ONS), is always very remarkable for gasoline/ethanol mixtures. Therefore, the prediction of their MONs remains a considerable challenge.

In this study, the auto-ignition properties of binary PRFs of 2,2,4-trimethylpentane andn-heptane with different RONs as well as PRF/ethanol blends with various ethanol contents were investigated in a constant-volume combustion chamber system. In addition, the relationship between the derived cetane number (DCN) and RON was examined. A close correlation was found between the ONS and the ignition delay sensitivity (IDS), which enabled the prediction of the RONs and MONs of the PRF/ethanol blends. Finally, two obtained equations relating the ONS with the IDS and the RON with the DCN were used to estimate the RONs and MONs of some ethanol-containing fuels.

2 Experimental

2.1 Test fuels

2,2,4-Trimethylpentane,n-heptane, and ethanol were purchased from Tianjin Damao Chemical Reagent Factory. Their purities, chemical formulas, and molecular weights are listed in Table 1. The compositional details,octane test results, and cetane test results of the test fuels are summarized in Table 2.

Table 1 Basic parameters of the fuel components

2.2 Experimental equipment

The spray auto-ignition properties were evaluated in a constant-volume combustion chamber with direct fuel injection into heated and compressed synthetic dry air[19].A dynamic pressure sensor was positioned to monitor the pressure wave generated during the combustion of the fuel. From the dynamic pressure curve, the ignition delay (ID) was determined as the period of time between the start of fuel injection and the onset of the initial pressure increase, which was defined as the point where the pressure exceeded the initial pressure plus 0.02 MPa[20]. The DCN was calculated according to the standard method NB/SH/T 6035[19]. In this work, the temperature under DCN standard conditions was 873.15K, which was assigned to HIGH T, and the ID was termed IDh. Furthermore, 853.15 and 833.15 K were selected as additional test temperatures, among which 833.15 K was assigned to LOW T, and the ID was termed as IDl. The initial chamber pressure, fuel pressure, coolant temperature, and injection time are listed in Table 3.

Table 2 Composition and properties of the test fuels.

Table 3 Experimental conditions in the constant-volume combustion chamber.

3 Results and Discussion

3.1 Auto-ignition of PRFs

Dynamic chamber pressure curves for the PRFs with different RONs measured at 873.15 K are presented in Figure 1. With the gradual increase of the knock resistance, the ignition delay time became longer and the overall combustion process was delayed. The detailed ID values are listed in Table 4. The rate of heat release (RHR)in these spray auto-ignition experiments can be expressed as follows[20]:

Table 4 ID values of PRFs with different RONs measured at 873.15, 853.15, and 833.15 K

whereQis the heat released,Pis the chamber pressure,V0is the chamber volume (473 mL), andγis the specific heat ratio (taken to be a constant value of 1.35 for simplicity).The RHR curves of the PRFs are plotted in Figure 2. It can be clearly seen that all of the PRFs displayed twostage heat release. The first stage of heat release might correspond to a kind of pre-reaction phenomenon during which some unstable active radicals accumulated to ignite a cool flame. However, the generated heat was not sufficient to lead directly to combustion.

Figure 1 Chamber pressure curves for PRFs with different RONs measured at 873.15 K and an initial pressure of 2 MPa

Figure 2 RHR curves for PRFs with different RONs measured at 873.15 K and an initial pressure of 2 MPa

Reducing the initial temperature of the combustion chamber would be expected to affect the ID values owing to the influence of temperature on fuel reactivity. Figure 3 shows the variation of the ID values with temperature based on the data listed in Table 4. In the narrow temperature range from 873.15 to 833.15 K, the PRFs did not exhibit the negative temperature coefficient phenomenon. The ID values increased with decreasing temperature for all of the PRFs. In addition, the trends were almost the same for the five PRFs, as indicated by the almost parallel curves. This suggests similar changes in the combustion kinetics with temperature for the PRFs. It was also the reason why PRF was selected as the reference fuel for calibration during the octane number testing.

Because both the cetane number and octane number characterize the activity of the auto-ignition reaction,there should exist a correlation between the RON and DCN. The PRFs were found to exhibit an excellent linear relationship between the RON and DCN, as shown in Figure 4 and expressed in Equation (2). From this perspective, although the calibration fuels used in the RON and DCN tests were different, both parameters reflected the auto-ignition behavior of the fuels, such that the RON and DCN were mutually predictable.

Figure 3 Variation of ID values of PRFs with different RONs measured at 873.15, 853.15, and 833.15 K

Figure 4 Relationship between RON and DCN for PRFs with different RONs

3.2 Auto-ignition of PRF/ethanol blends

Ethanol addition was expected to improve the anti-knock performance, which would be reflected in an extension of the combustion retardation period. Figure 5 presents the dynamic chamber pressure curves for the PRF/ethanol blends with various ethanol contents tested at 833.15 K. The PRF/ethanol blends were prepared based on PRF 90 with ethanol contents ranging from 2% to 10%. The corresponding RHR curves and ID values are shown in Figure 6 and Table 5,respectively. Ethanol was previously reported to lack lowtemperature ignition reactivity and favor chain-termination pathways forming acetaldehyde and hydroperoxy radicals over conventional chain-branching pathways forming hydroxyl radicals[3]. Meanwhile, the high latent heat of vaporization of ethanol reduces the charge-gas temperature when directly injected, which in turn delays the auto-ignition process[18]. Furthermore, a higher ethanol content leads to a lower heat value of the fuel. Therefore, the total heat release decreased with increasing ethanol content, as shown in Figure 6. In addition, the peak pressure of combustion declined with increasing ethanol content, as shown in Figure 5, which indicated a loss of power performance for the ethanol/gasoline blends.

Table 5 ID values of PRF/ethanol blends with various ethanol contents measured at 873.15, 853.15, and 833.15 K.

Table 6 IDS and ONS values of PRF/ethanol blends 90+Ex

Compared with the combustion characteristics observed for the PRFs (Figure 3), the PRF/ethanol blends exhibited distinct temperature-dependent behavior. As shown in Figure 7, the ID curves were no longer parallel but instead became steeper with increasing ethanol content. This implies that the chemical kinetics of the PRF/ethanol blends at low temperature were very different from those of the PRFs. Furthermore, the difference was magnified with increasing ethanol content. The low-temperature characteristics were ascribed to the special physical and chemical properties of ethanol. As shown in Figure 8, PRF 100 and PRF/ethanol blend 90+E10 displayed similar ID values under DCN conditions. However, they exhibited distinct ID behavior at low temperature. This was also the case for PRF 98 and PRF/ethanol blend 90+E5 and for PRF 95 and PRF/ethanol blend 88+E5. PRF/ethanol blends 88+E5 and 90+E5 had the same ethanol content and appeared to show the same temperature-dependent behavior, as demonstrated by the parallel ID curves. The difference in the ID variation with temperature for these test fuels was related to the IDS and octane number sensitivity(ONS), which is discussed in the following section.

The relationship between the RON and DCN for the PRF/ethanol blends 90+Ex(wherexrepresents the percent ethanol content) was examined as shown in Figure 9,revealing a non-linear correlation unlike that observed for the PRFs. The obtained fitting curve was as follows:

Figure 5 Chamber pressure curves for PRF/ethanol blends with various ethanol contents measured at 833.15 K and an initial pressure of 2 MPa

Figure 6 RHR curves for PRF/ethanol blends with various ethanol contents measured at 833.15 K and an initial pressure of 2 MPa

Figure 7 Variation of ID values of PRF/ethanol blends with various ethanol contents measured at 873.15, 853.15, and 833.15 K

Figure 8 Variation of ID values of selected PRFs and PRF/ethanol blends measured at 873.15, 853.15, and 833.15 K

Figure 9 Relationship between RON and DCN for PRF/ethanol blends

This equation enables the RONs of gasoline/ethanol blends to be predicted solely on the basis of their DCN test results, which saves time and fuel compared with RON testing.

3.3 Relationship between IDS and ONS and estimation of RON and MON

On the basis of the above experimental results and analysis, an IDS concept for gasoline/ethanol blends was proposed[21]with the following form:

where IDl,PRF/ethanoland IDl,PRFdenote the ignition delay times measured at 833.15 K for a PRF/ethanol blend and the corresponding PRF, respectively. Specifically,the ID value of a given PRF/ethanol blend under DCN conditions should be approximately equal to that of the PRF. Therefore, it was necessary to establish the relationship between IDland IDhfor the PRFs, because it was impossible to obtain all of the ID changes with temperature for PRFs with different RONs. As demonstrated by the above results, the five studied PRFs displayed parallel ID curves, which meant that the PRFs exhibited an IDS value of 1. As shown in Figure 10, a relatively good correlation between IDland IDhwas observed for the PRFs. The resulting relationship is expressed in Equation (5), which allows the IDS of a given PRF/ethanol blend to be easily determined by only measuring its ID values at 833.15 and 873.15 K.

Figure 10 Relationship between IDl and IDh for the PRFs

Combining Equations (4) and (5) allows the IDS values of the PRF/ethanol blends to be calculated. Figure 11 shows the variation of IDS and DCN with the ethanol content for the PRF/ethanol blends 90+Ex. As the ethanol content was increased from 0% to 10%, the IDS increased from 1.000 to 1.227, with a linear region up to an ethanol content of 8% followed by a steep rise. Similar behavior was observed for the DCN, which switched from a linear dependence on the ethanol content to a slower decrease beyond an ethanol content of 8%, although the effect was more pronounced for the IDS.

Figure 11 Variation of IDS and DCN with the ethanol content for the PRF/ethanol blends 90+Ex

Figure 12 Correlation between the ONS and IDS values for the PRF/ethanol blends 90+Ex

A comparison of the IDS and ONS values for the PRF/ethanol blends 90+Exis presented in Figure 12, and the detailed values are listed in Table 6. A logarithmic correlation between the ONS and IDS was observed withR2= 0.9402, as expressed in Equation (6). This equation allows the ONS of a given gasoline/ethanol fuel to be estimated by determining its IDS.Using the concept described above, the RON and MON values of four ethanol-containing fuels were estimated and compared with the values measured in engine tests.The results are presented in Table 7. As an example, the IDland IDhvalues of 92+E5 were obtained from constantvolume combustion chamber experiments performed at 833.15 and 873.15 K (DCN conditions), respectively.The corresponding IDlof the PRF was calculated using Equation (5) based on the assumption that IDh,92+E5≈IDh,PRF. Furthermore, the IDS and ONS values of the PRF/ethanol blend 92+E5 were determined according to Equations (4) and (6). As the constant-volume combustion chamber experiment provided the DCN value, combined with Equation (3) and the ONS value, the RON and MON values could be estimated without the need to conduct complex and tedious engine tests.

Table 7 Estimation of RON and MON values for four gasoline/ethanol fuels

As observed in Table 7, the estimation accuracy for RON and MON was relatively high when the ethanol content was less than 8%, where the minimum error between the predicted and experimental RON values was 0.3 and the minimum error between the predicted and experimental MON values was ?0.1. However, the accuracy gradually decreased with increasing ethanol content. For PRF/ethanol blend 92+E10, the difference could reach two to three units. Finally, the practical gasoline/ethanol blend E5 was selected to evaluate the feasibility of the proposed method. There was also a certain deviation between the predicted value and the measured value.This was primarily attributable to the complexity of the components in the practical gasoline/ethanol blend, where aromatic hydrocarbon compounds with a higher octane sensitivity exerted a marked influence on the ONS. In spite of this, the proposed method indeed provided a means to predict the octane number, and for PRF/ethanol blends with lower ethanol contents, the prediction results were relatively good.

4 Conclusions

In this work, the spray auto-ignition properties of binary PRFs of 2,2,4-trimethylpentane andn-heptane with different RONs as well as PRF/ethanol blends with various ethanol contents were investigated in a constantvolume combustion chamber system taking the NB/SH/T 6035 standard as a reference. The five studied PRFs displayed parallel temperature-dependent ID curves,indicating similar auto-ignition properties, an IDS of one,and an ONS of zero. The presence of ethanol exerted a marked influence on the auto-ignition properties. In contrast to the PRFs, the ID values of the PRF/ethanol blends increased sharply with decreasing temperature.Therefore, the IDS and ONS values were greater. The relationships between the RON and DCN values and the ONS and IDS values enabled the prediction of both RON and MON, although their accuracy has yet to be confirmed. For practical gasoline/ethanol blends, the interactions between the various components, especially aromatic compounds, must be considered. The results of this work provide guidance for the selection of standard materials for the determination of the octane numbers of gasoline/ethanol mixtures in engine tests. Furthermore,the obtained combustion-related data give important information for understanding the chemical kinetics of gasoline/ethanol mixtures.

Acknowledgment:This work was funded by the National Key Research and Development Program (2017YFB0306505).