- Email: [email protected]
Relating the octane numbers of fuels to ignition delay times measured in an ignition quality tester (IQT)
Fuel 187 (2017) 117–127
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Full Length Article
Relating the octane numbers of fuels to ignition delay times measured in an ignition quality tester (IQT) Nimal Naser a,⇑, Seung Yeon Yang a, Gautam Kalghatgi b, Suk Ho Chung a,⇑ a b
King Abdullah University of Science and Technology (KAUST), Clean Combustion Research Center (CCRC), Thuwal, Saudi Arabia Saudi Aramco Research and Development Center, Fuel Technology Research and Development Division, Dhahran, Saudi Arabia
h i g h l i g h t s A methodology for estimating the RON and the MON of fuels using an IQT is presented. Estimated RON and MON are in good agreement with values obtained from standard tests. The DCN of PRFs, TRFs, TPRFs FACE and certification gasolines are reported. Correlation between DCN and RON is provided.
a r t i c l e
i n f o
Article history: Received 5 May 2016 Received in revised form 2 September 2016 Accepted 4 September 2016
Keywords: Octane index Octane numbers Ignition quality tester Autoignition Combustion Gasoline Derived cetane number
a b s t r a c t A methodology for estimating the octane index (OI), the research octane number (RON) and the motor octane number (MON) using ignition delay times from a constant volume combustion chamber with liquid fuel injection is proposed by adopting an ignition quality tester. A baseline data of ignition delay times were determined using an ignition quality tester at a charge pressure of 21.3 bar between 770 and 850 K and an equivalence ratio of 0.7 for various primary reference fuels (PRFs, mixtures of isooctane and n-heptane). Our methodology was developed using ignition delay times for toluene reference fuels (mixtures of toluene and n-heptane). A correlation between the OI and the ignition delay time at the initial charge temperature enabled the OI of non-PRFs to be predicted at specified temperatures. The methodology was validated using ignition delay times for toluene primary reference fuels (ternary mixtures of toluene, iso-octane, and n-heptane), fuels for advanced combustion engines (FACE) gasolines, and certification gasolines. Using this methodology, the RON, the MON, and the octane sensitivity were estimated in agreement with values obtained from standard test methods. A correlation between derived cetane number and RON is also provided. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Engine knock, an abnormal combustion phenomenon that occurs in internal combustion (IC) engines, limits the thermal efficiency of spark ignition (SI) engines [1]. It is caused by autoignition in hot spots at the unburned end gas ahead of the expanding flame front and depends on pressure and temperature evolution in the end gas as well as on the anti-knock quality of the fuel. The autoignition or anti-knock quality of a fuel is measured by the research octane number (RON) and the motor octane number (MON), which are determined using the standard test methods prescribed by ASTM D2699 and D2700 [2,3], respectively, with a ⇑ Corresponding authors. E-mail addresses: [email protected] (N. Naser), [email protected]. sa (S.H. Chung). http://dx.doi.org/10.1016/j.fuel.2016.09.013 0016-2361/Ó 2016 Elsevier Ltd. All rights reserved.
cooperative fuel research (CFR) engine. In these tests, the knock behavior of a fuel is compared to that of primary reference fuels (PRFs), mixtures of iso-octane and n-heptane, with known isooctane volume percentages. The RON (or the MON) of the test fuel can then be related to that of the PRF that matches its knock behavior. There are, however, differences in the chemical composition and autoignition chemistry of practical fuels and the PRFs that are used to define the octane scale. In addition, the same gasoline may match different PRFs at different operating conditions. For all practical fuels, the RON is higher than the MON because the temperature of an unburned gas is higher at a given pressure under MON than under RON conditions. This is due to the more severe MON testing conditions, which are performed at an intake mixture temperature of 422 K and an engine speed of 900 rpm as compared with RON testing conditions, which are performed at 325 K and
118
N. Naser et al. / Fuel 187 (2017) 117–127
600 rpm. As a result, in MON testing, the autoignition chemistry of PRFs is dominated by negative temperature coefficient (NTC) behavior [4]. Practical fuels, at similar conditions, however, do not exhibit NTC behavior and therefore demonstrate the antiknock characteristics of a lower-octane-quality PRF than what was used in RON testing: RON values are higher than MON values [4]. The anti-knock quality of a fuel can best be described by the octane index (OI) [1,5–8], which is defined as
OI ¼ ð1 kÞRON þ kMON
ð1aÞ
OI ¼ RON kS
ð1bÞ
Here, the weighting factor k is a constant depending on pressure/ temperature histories with time in an end gas [1] and engine speed, and is weakly dependent on mixture strength [1,9]; S is the octane sensitivity defined as (RONMON); and octane index (OI) is the volume percent of iso-octane in the PRF that matches the knock behavior of the fuel at any given condition. By definition, k = 1 under MON testing conditions and k = 0 under RON testing conditions. If the temperature of an unburned gas for a specified pressure decreases, the k value decreases [1,6]. Sensitivity (S) can be regarded as a measure of the difference in autoignition chemistry of a fuel at the two operating conditions, while k is a measure of how different the operating conditions are between the two tests. The temperature of the unburned bulk gas at 15 bar designated as Tcomp15 was introduced in [6], and this information was used to establish a given operating condition with respect to the RON test condition. Empirically, it has been demonstrated that the k value decreases as Tcomp15 decreases [1,6]. SI engine design has developed to enable higher pressure for a given temperature to achieve better thermal efficiency and power density. So historically, Tcomp15 has decreased from the value representative of the MON test such that the importance of MON in describing fuel anti-knock quality has decreased [1,9]. In fact, Tcomp15 in modern engines is even lower than in the RON test condition and k can be negative, and thus for a given RON, a fuel with lower MON has higher OI and is more resistant to knock. Further efforts to improve SI engine efficiency by downsizing and turbocharging are driving the trend towards more negative k values, while old engines and CFR engines have positive k values [9]. Attempts have been made to interpret the RON and the MON of practical fuels in terms of more fundamental measures of autoignition, such as ignition delay time. Bradley et al. [10] showed that ignition delay times of non-PRFs can be deduced reasonably from the OIs measured in a variety of engine tests using the data available on ignition delay times for PRFs. Analogous to the k weighting factor relating OI to RON and MON, they defined the j factor relating the ignition delay times of non-PRFs to those for PRFs under the same conditions. Perez and Boehman [11] measured the ignition delay times in an ignition quality tester (IQT) for 21 surrogate gasoline fuels to find a poor correlation between the measured ignition delay time with the RON and the MON. Kalghatgi et al. [12] suggested that fuels should be rated on a toluene/n-heptane scale rather than using PRFs because with toluene/n-heptane blends, toluene reference fuels (TRF), the sensitivity will be high and comparable to that of practical fuels. Using the current RON test, a fuel could be assigned a toluene number by the volume percent of toluene in the TRF blend matching the fuel behavior in the RON test. However, it is possible to find a ternary mixture of toluene, iso-octane, and n-heptane (toluene/PRF or toluene primary reference fuel (TPRF) blends) that matches both the RON and the MON of a fuel. Kalghatgi et al. [13] developed empirical correlations to predict the RON, the MON, and octane sensitivity for TPRF blends based on experimental RON/MON measurements of such mixtures.
Standard RON and MON tests are time consuming and require strict control of various parameters involved during engine operation. The methodology presented in this paper, which can be a convenient initial screening tool to determine RON and MON, estimates the RON and the MON from a combustion device other than the CFR engine. This methodology was developed using a constant volume combustion chamber with liquid fuel spray injection, which is capable of operating at various temperatures and pressures. For wide applicability of the developed methodology, a constant volume combustion chamber of a standard IQT (Advanced Engine Technology Ltd.) system was adopted, which is used to determine the derived cetane number (DCN) of diesel fuels based on ASTM D6890 [14] at the charge temperature and pressure of 818 ± 30 K and 21.37 ± 0.07 bar, respectively. The ignition delay time, sid, was measured as the time between the start of fuel injection and the start of significant heat release; an average of 32 firings were taken. The DCN is related to the ignition delay time in [ms] by the formula DCN = 4.46 + (186.6/sid) for ignition delay times between 3.1 and 6.5 ms, which correspond to DCNs between 64 and 33. The KAUST research IQT (KR-IQT) was modified from the standard IQT, which typically has a fixed volume of injected fuel. Since equivalence ratio is an important parameter affecting ignition delay times, we have made modifications to testing methods adopted in this study as described in [15] to include this ratio (details in the next section). Ignition delay times for fuels in the diesel autoignition range are typically short and thus contain both physical (evaporation and mixing) and chemical (reactive) time contributions. For fuels in the gasoline autoignition range, RON > 60 and DCN < 30 [6], with relatively long ignition delay times, the physical delay may be much shorter than the chemical delay, which is one motivation for using the KR-IQT for determining OI. Ignition delay times for the fuel mixtures of PRFs were measured first to provide a baseline data set. Ignition delay time data for octane sensitive TRFs were then used to develop a methodology. Fuel mixtures of toluene primary reference fuels (TPRFs), fuels for advanced combustion engines (FACEs), and certification gasolines were used to validate the methodology.
2. Experiment A schematic of the experimental setup is shown in Fig. 1. The apparatus comprises a KR-IQT, a gas supply system, and several data acquisition setups. The KR-IQT is equipped with a constant volume combustion chamber at 0.21 L, filled with an oxidizing medium that can be pressurized up to a maximum of 21.6 bar and heated with an electric heater to temperatures in the range of 670–860 K. A liquid fuel spray is injected into a pressurized and heated oxidizing environment. A pneumatically driven mechanical pump delivers fuel stored in a fuel reservoir to the injector with a single-hole S-type inward opening pintle nozzle [16]. The temperature of the nozzle tip during operation is around 325 K. The injected fuel mass can be regulated by adopting a pneumatically driven variable displacement pump (VDP), which varies displacement of the pump, thereby the volume of fuel injected. This modification was provided by the manufacturer. It uses turnbuckle bolts, which when rotated, change the displacement between the front body and rear body of VDP using a chain system as was shown in Fig. 3 in [15]. The VDP has a limited displacement range, therefore, for a specified charge pressure, there is a limited range of equivalence ratio available depending on the density and oxygen requirement of the fuel. More details on mass calibration for the VDP have been explained in detail in [15,17]. The fuel is injected into the combustion chamber at a pressure of 225 bar [16]. A piezo-electric pressure transducer mounted
N. Naser et al. / Fuel 187 (2017) 117–127
119
Fig. 1. Schematic of the experimental setup.
along the same axis as that of the fuel injector acquires a chamber pressure signal, which is collected by a data acquisition system along with the needle lift values from a needle lift sensor placed in front of the injector. These two signals form the backbone of measuring ignition delay times of test fuels. A total ignition delay time, sid , includes delays due to both physical and chemical processes during ignition. Physical ignition delay time describes the time required for the injected liquid spray to atomize into fine droplets, heat, evaporate, and mix with oxidizer to form a fuel-oxidizer mixture that can ignite. Chemical ignition delay time describes the time required for fuel molecules to breakdown into smaller fragments and react with oxidizer molecules to form sufficient chain branching radicals that ignite the mixture [18]. The ignition delay times presented throughout this paper are total ignition delay times. As mentioned previously, ignition delay time is relatively much longer for gasoline fuels than for diesel fuels such that the contribution from a physical process is expected to be relatively small. The total ignition delay time is defined as the time difference between the start of injection (SoInj) and the start of ignition (SoIgn), as indicated in Fig. 2. The SoInj was set as the point where the needle lift attains its maximum. The SoIgn is determined differently from the conventional IQT system software and modified definitions are used [15]. In the conventional IQT software, SoIgn is determined as the point where the difference in chamber pressure P and the initial charge pressure P0 of DP exceeds a preset value of 1.38 bar. This definition is suitable for diesel-like fuels, where distinctive low-temperature heat release is not conspicuous due to shorter ignition time scales. This definition is, however, unsuitable for fuels that exhibit low-temperature heat-release behavior because the point of ignition is determined to be much earlier than the actual point of ignition, as demonstrated in Fig. 2. Here, the chamber pressure is shown as a function of time t for iso-octane fuel, and the conventional IQT definition, represented by the vertical dotted line marked as SoIgn (IQT), is shown to be unable to correctly represent the rapid pressure rise near
Fig. 2. Typical change in pressure with time of iso-octane ignition at P0 = 21.3 bar, T0 = 720 K, and / = 0.7, indicating various definitions of SoIgn.
t = 37 ms. Note that this delay time is much longer than the previously mentioned typical range of the DCN test of 3.1–6.5 ms. To compensate for the effect of the low-temperature heat-releaseinduced pressure rise, Bogin et al. [19] used a preset value of 2.76 bar with this same criteria, represented by the vertical line as SoIgn (Bogin et al.), but they failed to correctly predict rapid pressure rises. Instead, the gradient method proposed in [15] is used to accurately determine the start of ignition while allowing for the effects of low-temperature heat release. This method determines the point of ignition as the intersection of two line tangents (LT) in a pressure–time (P-t) diagram. The first line tangent (dP/dt) (LT 1) is determined at the point where the chamber regains its initial pressure. As shown in Fig. 2, the pressure initially decreases due to evaporative cooling of the liquid fuel spray droplets, but as the chemical reaction proceeds, the pressure returns to the
120
N. Naser et al. / Fuel 187 (2017) 117–127
Fig. 3. Ignition delay times of PRFs.
initial level. The second line tangent (LT 2) is selected as the maximum slope of the pressure rise during ignition. The point where these two line tangents intersect is designated as the SoIgn (gradient method). This method successfully captured the timing of rapid pressure rise. All fuel samples were measured at P0 = 21.3 bar with an equivalence ratio of / = 0.7, a pressure required by the DCN standard ASTM D6890 [14]. Rather than fixing the volume of the injected fuel as in determining DCN, we fixed the equivalence ratio. This approach was successful because the VDP accurately evaluated the equivalence ratio window for each fuel despite differences in fuel densities and stoichiometric air-fuel ratios. Selection of the equivalence ratio at / = 0.7 was based on the ability of this value to represent a gasoline compression ignition condition for advanced engines with the lean-burn strategy [20]. Since many gasoline engines are operated at stoichiometric conditions, cases using an equivalence ratio of / = 1.0 were also tested for certain fuels and the results are presented in the Supplementary Material. We selected an initial pressure of P0 = 15 bar, motivated by the definition of Tcomp15 in [6], relating the temperature of mixture at 15 bar during engine experiment to RON and MON values. This pressure is therefore likely ideal for correlating IQT conditions to IC engine conditions during stoichiometric engine operations. As mentioned previously, this pressure condition satisfies the limitations on the displacement range of VDP at / = 1.0. The charge temperature is varied from 770 to 850 K, which, for most modern engines, is a temperature range that corresponds to compression temperatures [21]. The zero air oxidizer was used for all tests (air containing trace amounts of total hydrocarbons <0.1 ppm) with an O2 concentration of 20.9 ± 1% balanced with nitrogen. All experiments are conducted following the standard procedure specified in ASTM D6890. For each test condition, 15 pre-injections are carried out to establish a stable, uniform, steady environment for the 32 main injections that follow. The data of these main injections were used for processing. A MATLAB code was developed to accommodate the new definition of SoIgn we described. The raw chamber pressure signal is smoothed or filtered to eliminate fluctuations in the signal due to resonance and noise, as can be seen in Fig. 2. Although severe pressure oscillations were observed mainly after the rapid pressure rise in Fig. 2, there were also non-negligible pressure fluctuations during the induction period. Zheng et al. [22] employed an eleven-point moving average and a finite impulse response filter with phase correction to remove some important frequency com-
ponents; however, these adjustments failed to satisfactorily eliminate the frequency of some of the components that had to be removed. Thus, the data processing code comprised a local regression using weighted linear least squares and a second-degree polynomial model (locally weighted scatterplot smoothing, LOESS) [23]. Details on data processing are available in [15]. Here, ignition delay time was determined for four sets of fuel samples, as specified in Tables 1 and 2. The first set comprised PRFs, whereby PRF X indicates a mixture of an X volume percentage of iso-octane and the remaining (100 X) volume percentage of n-heptane to provide a baseline data set. The second set comprised TRFs, whereby TRF X contains an X volume percentage of toluene. The third set comprised TPRFs, where TPRF X-Y indicates a mixture of PRF X with a Y volume percentage of toluene, and the fourth set comprised FACE gasolines [24] and the certification gasoline fuels listed in Table 2 (collectively called real fuels). The RON and the MON of TRFs and TPRFs were obtained from the standard ASTM tests conducted in [12,13], and when experimental octane numbers were not available, they were calculated with empirical correlations developed by Kalghatgi et al. [13]. The RON and the MON of the FACE and the certification gasolines were obtained from the certificate of analysis provided by the fuel suppliers. All fuels selected for this study had RON/MON values greater than 40 because the standard method for determining octane number from a CFR engine have a digital counter setting for octane numbers above 40 and hence can only determine the octane number for fuels that have RON/MON values above 40.
3. Methodology for estimating RON, MON, and OI 3.1. Baseline data for PRFs and TRFs We sought to answer whether ignition delay times obtained from IQT experiments could be used to estimate RON, MON, and OI. Thus, we made the assumption that fuels with the same octane index operating under the same pressure/temperature history have the same ignition delay time [12]. We base this basic premise of the methodology on understanding the variation of ignition delay times of a non-PRF with respect to PRFs at different temperatures. Depending on the sensitivity of a fuel, a non-PRF can match different PRFs at different temperatures if it has considerable sensitivity. In such a case, best fit temperatures from OI data will be selected, which can estimate RON/MON values. Note that a nonPRF with zero sensitivity can match a particular PRF at all temperatures. From this assumption, we aimed to find the OI of non-PRFs at various temperatures while maintaining all other parameters constant, including charge pressure, equivalence ratio, and injection pressure. First, the ignition delay times of PRFs were measured using the KR-IQT to determine the relationship between OI and ignition delay times at various initial temperatures at 21.3 bar and / = 0.7. The ignition delay times of PRFs against the reciprocal of charge temperature 1000/T0 are shown in Fig. 3. Results showed that the logarithm of the total ignition delay increases as the charge temperature (T0) decreases in a reasonably linear manner, indicating an Arrhenius-type temperature dependence; ignition delay time also increases with PRF X. The ignition delay times of non-PRFs are also tested at the same initial conditions and the results for the TRF fuels are shown in Fig. 4 and the PRF data from Fig. 3 are also shown with dotted lines. We observed that TRFs show a similar trend to that for PRFs: the logarithm of ignition delay time increases linearly with decreasing charge temperature. Note that a fuel that has zero or little sensitivity has an ignition delay curve that is nearly parallel to the corresponding PRF, as will be shown later when we validate our
121
N. Naser et al. / Fuel 187 (2017) 117–127 Table 1 Fuel matrix of PRFs, TRFs, and TPRFs. Fuel
*
PRF
Toluene vol.%
Vol.%
MON
RON/MON Refs.
Toluene
iso-octane
n-heptane
PRF PRF PRF PRF PRF PRF PRF
0 50 60 70 80 90 100
0 50 60 70 80 90 100
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 50 60 70 80 90 100
100 50 40 30 20 10 0
0.0 50.0 60.0 70.0 80.0 90.0 100.0
0.0 50.0 60.0 70.0 80.0 90.0 100.0
[2,3] [2,3] [2,3] [2,3] [2,3] [2,3] [2,3]
TRF TRF TRF TRF TRF TRF TRF
40 45 50 55 60 65 70
0 0 0 0 0 0 0
40 45 50 55 60 65 70
40 45 50 55 60 65 70
0 0 0 0 0 0 0
60 55 50 45 40 35 30
53.7 59.7 65.6 71.3 77.9 84.5 89.3
47.1 52.4 57.6 62.6 68.5 74.5 78.2
BR* [13] BR [13] [12] BR [13] [12] [12] [12]
TPRF TPRF TPRF TPRF
60-10 60-20 60-30 60-40
60 60 60 60
10 20 30 40
10 20 30 40
54 48 42 36
36 32 28 24
66.0 73.6 79.0 86.2
64.4 70.0 74.0 79.6
[13] [13] [13] [13]
TPRF TPRF TPRF TPRF
70-10 70-20 70-30 70-40
70 70 70 70
10 20 30 40
10 20 30 40
63 56 49 42
27 24 21 18
74.8 80.7 86.1 91.0
72.8 76.9 80.6 84.0
BR BR BR BR
TPRF TPRF TPRF TPRF
80-10 80-20 80-30 80-40
80 80 80 80
10 20 30 40
10 20 30 40
72 64 56 48
18 16 14 12
84.5 89.1 92.8 96.7
82.0 85.6 86.9 88.7
[13] [13] [13] [13]
TPRF 90-10 TPRF 90-20 TPRF 90-30
90 90 90
10 20 30
10 20 30
81 72 63
9 8 7
92.0 95.3 98.3
90.0 91.4 92.7
BR [13] BR [13] BR [13]
[13] [13] [13] [13]
BR indicates blending rule.
Table 2 Fuel matrix of FACE and certification gasolines.
*
RON
Fuel
RON
MON
Ref.
FACE A FACE F FACE I FACE J Coryton gasoline Haltermann gasoline
83.5 94.4 70.3 71.8 97.5 91.0
83.6 85.8 69.6 68.8 86.6 83.4
COA* COA COA COA COA COA
COA indicates certificate of analysis.
methodology with real fuels having zero or low sensitivity. Note that although the slopes of all TRFs appear to be similar, this is an artifact created by plotting them on a logarithmic scale. To correct for this artifact, we plotted this data on a linear scale, which highlights the difference in their slopes (Fig. S6 in Supplementary Material). Similarly, a linear scale shows the differences in the slopes of PRFs are (Fig. S6 in Supplementary Material). As the sensitivity of the fuel increases, the ignition delay time curves intersect with the PRF X curve. In other words, the difference in slope between ignition delay time curve of an unknown fuel and the adjacent PRF curves is related to the fuel’s octane sensitivity. The TRF data exhibits this behavior because toluene is a component that influences sensitivity to the resulting mixture. 3.2. Correlation with OI Known OI values for PRFs are plotted as a function of ignition delay times in Fig. 5. Note that the RON and the MON values of PRFs are the same by definition and their sensitivities are 0; as a result OI = RON = MON. Thus, for a specified PRF, OI does not vary at different charge temperatures. For a specified temperature, there is a gradual reduction in the slope [d(OI)/dsid] as ignition delay time increases. For a fixed charge temperature, OI can be related to ignition delay time by the following formula:
OI ¼ a½ln sid þ b: n
Fig. 4. Ignition delay times of TRFs.
ð2Þ
As will be shown later in this section, data representation in the functional form of Eq. (2) has a linear relationship with temperature in the OI against the [lnsid]n plot. The exponent n was selected by an ordinary least-squares approach by creating a loop in a range of 10 6 n 6 10 with a step
122
N. Naser et al. / Fuel 187 (2017) 117–127
Fig. 5. OI of PRFs at several temperatures.
Fig. 7. OI of TRFs.
of 0.1 to obtain maximum coefficients of determination. The maximum coefficient of determination R2 is obtained when n = 2.4. Fig. 6 shows that the expected linear relationship between OI n and ½lnsid assumed in Eq. (2) does indeed hold, evidenced by the linear isotherms. The best fit parameters of a and b are both dependent on temperature, and they have the quadratic variation as given by the following equations (comparison with other regression models are shown in Fig. S7 in Supplementary Material):
for a given T0 (isotherm) and the measured value ignition delay time (vertical line from the horizontal axis). Fig. 7 shows the estimates of OIs of TRFs at several temperatures determined in this manner. The OI values of all TRFs decrease with increasing temperature. The OIs of TRFs with higher toluene content are observed to
The OI of a TRF can be determined using the measured ignition delay time with the initial charge temperature. This assumption is based on the premise that for the same ignition delay time under the same operating conditions, non-PRFs and PRFs have the same OI. On an OI versus [ln sid]2.4 plot with the same isotherms as those in Fig. 6, OI for a TRF can be determined from the cross point
decrease more rapidly with ½lnsid for the same temperature change, which is an indication of the octane sensitivity of the fuels. Alternatively, the OI of TRFs varies with ignition delay time as temperature varies. The slope for the specified charge temperature in Fig. 7 is a qualitative indication of the k value given in Eq. (1). As temperature increases, the absolute value of the slope of the isothermal line decreases, indicating a smaller change in OI with change in ignition delay. At relatively lower temperatures, the slope increases where the change in OI with ignition delay is higher. Note that the isotherms intersect for OI values greater than about 100, but this cross-over of isotherms does not pose a problem in determining estimated RON and MON values, as we will demonstrate next. If the OI from Fig. 7 is plotted against the charge temperature, there is reasonably linear variation with the temperature for each TRF, as shown in Fig. 8. The decrease in OI or autoignition quality of
Fig. 6. OI versus [ln sid]2.4 of PRFs exhibiting linear iso-T0 lines (values in parentheses indicate R2).
Fig. 8. OI versus charge temperature demonstrating the extrapolation method to estimate RON and MON from IQT data for TRFs.
2 T0 T0 a ¼ 39862 þ 70652 31456 1000 1000
ð3aÞ
2 T0 T0 2216 þ 1190 b ¼ 1121 1000 1000
ð3bÞ
2:4
123
N. Naser et al. / Fuel 187 (2017) 117–127
values from the literature designated as RONnominal and MONnominal in Fig. 9. The nominal values were obtained using the standard tests [2,3] obtained in [12,13] and by the correlations in [13], which are also listed in Table 3 together with the OIT values. These results demonstrate that the estimated OIT values are in good agreement with the nominal values as shown by the correlation coefficients of R2 = 0.965 and 0.878 for RON and MON, respectively.
the fuel with increasing charge temperature is evident. The method of determining the RON and the MON from these data is described as follows. As can be observed in Fig. 8, the OI values (solid symbols) of all the fuels are higher than their corresponding nominal RON and MON values given in Table 1. A linear extrapolation of the OI curve for each fuel (solid lines) to temperatures beyond the operating temperature range of KR-IQT indicates that OI values decrease at higher temperatures. A linear regression fit of each OI curve and a substitution of RON and MON values provides the temperatures at which the estimated OI values match nominal RON and MON values. Note that by adjusting the temperature for each fuel, the temperature at which the OI matches the RON/MON value exactly can be identified, although it could be slightly different for different fuels; however, temperatures are always similar, leading to small differences in estimated RON and MON values if this common fixed temperature is used. In line with the above discussion, two common fixed octane index temperatures (OIT) that provide best estimates of RON and MON values are identified at 875 K and 930 K for RON and MON, respectively. For an octane sensitive fuel such as a TRF, the OI decreases as temperature increases at a fixed pressure (i.e., it becomes more prone to autoignition and knock). In Fig. 8, the value of k (see Eq. (1a)) is 0 and 1 at 875 K and 930 K, respectively, and the slope of the line for each TRF depends on its octane sensitivity, S, which increases as the toluene concentration in the TRF increases. The estimated RON and MON values obtained with the OITs designated as RONOIT and MONOIT are compared with the nominal
3.3. Validation with TPRFs and real fuels To validate this method further, ignition delay times were measured for TPRFs and real fuels at four specified initial temperatures from the IQT experiments. Of the FACE gasolines, FACE A has the lowest octane number with very small sensitivity (S = 0.1) and FACE F has the highest octane number. FACEs I and J have a RON of about 70, and FACE I has a lower sensitivity. Both can be considered suitable fuels for gasoline compression ignition engines [25]. Figs. 10 and 11 show the ignition delay times with the inverse of the initial temperature for TPRFs and real fuels, respectively. To determine their OIs, the same methodology was applied to TPRFs and real fuels as was applied to TRFs. For the sake of clarity only selected TPRFs are shown in Figs. 10 and 12(b). Fig. 10 shows that as the volume percentage of toluene increases, the slopes of TPRFs gradually increase from the slopes of PRFs; for example, TPRF-60-10 (10%vol toluene) has an ignition delay time curve nearly identical to PRF 70, while TPRF-70-40 (40% vol toluene) has larger slope than PRF 90, indicating higher octane sensitivity. Similarly, in Fig. 11, low-octane sensitivity FACE I has a nearly identical slope to that of PRF 70. Meanwhile high-octane sensitivity
Fig. 9. Comparison of predicted and nominal values of RON (top) and MON (bottom) of TRFs.
Fig. 10. Ignition delay times of TPRFs.
Table 3 RON and MON values of TRFs from the OITs and the literature. OIT Fuel TRF TRF TRF TRF TRF TRF TRF
40 45 50 55 60 65 70
Nominal/literature
RON
MON
S
RON
MON
S
54.3 61.6 67.6 73.4 79.1 85.0 90.1
48.0 55.2 61.1 66.4 71.3 76.8 80.6
6.3 6.4 6.5 7.0 7.8 8.2 9.5
53.7 59.7 65.6 71.3 77.9 84.5 89.3
47.1 52.4 57.6 62.6 68.5 74.5 78.2
6.6 7.3 8.0 8.7 9.4 10 11.1
124
N. Naser et al. / Fuel 187 (2017) 117–127
temperature plots in Fig. 12. The data are shown in Tables 4 and 5 with the nominal values from [13] from the CFR tests and the data from the certificates of analysis for the TPRFs and real fuels, respectively. The near horizontal lines on the plot evidence the low sensitivity of both FACE A and I gasolines. The estimated RON and MON values match well with the nominal values, and this result is further exhibited in Fig. 13. The correlation coefficients are acceptable for the initial screening of TPRF and real fuels for the RON and the MON prediction from an IQT experiment.
3.4. Weighting factor k From the RON and the MON values obtained using the methodology described here, it is possible to determine k values given in Eq. (1) as a function of T0,
k¼ Fig. 11. Ignition delay times of real fuels.
Coryton gasoline has a larger slope than PRF 100, indicating a greater change in OI with temperature variation. This same methodology was used to estimate the RON and the MON for the TPRFs and the real fuels using the OI versus charge
RON OI RON MON
ð4Þ
The values of k obtained as a function of charge temperature are shown in Fig. 14 for all the fuels tested. The results show that k values are reasonably independent of the TRFs, the TPRFs, and the real fuels tested and that they depend on the physical operating condition of charge temperature, as per the definition of OI. The respective mean values, kmean, for the TRFs, the TPRFs, and the real fuels
(a)
(b)
Fig. 12. Predicted RONs and MONs of (a) TPRFs and (b) real fuels.
Table 4 RON and MON values of TPRFs from the OITs and the literature. OIT Fuel
Nominal/literature
RON
MON
S
RON
MON
S
TPRF TPRF TPRF TPRF
60-10 60-20 60-30 60-40
67.8 75.5 81.5 88.2
65.9 73.2 77.9 83.4
1.8 2.3 3.6 4.8
66.0 73.6 79.0 86.2
64.4 70.0 74.0 79.6
1.6 3.6 5.0 6.6
TPRF TPRF TPRF TPRF
70-10 70-20 70-30 70-40
76.5 82.4 87.5 92.2
75.3 80.7 85.0 85.1
1.2 1.7 2.5 7.0
74.8 80.7 86.1 91.0
72.8 76.9 80.6 84.0
2.0 3.8 5.5 7.0
TPRF TPRF TPRF TPRF
80-10 80-20 80-30 80-40
84.1 88.8 92.8 97.1
82.0 85.8 86.4 88.5
2.1 3.1 6.4 8.6
84.5 89.1 92.8 96.7
82.0 85.6 86.9 88.7
2.5 3.5 5.9 8.0
TPRF 90-10 TPRF 90-20 TPRF 90-30
92.7 95.6 98.5
90.4 89.7 89.8
2.3 5.9 8.7
92.0 95.3 98.3
90.0 91.4 92.7
2.0 3.9 5.6
125
N. Naser et al. / Fuel 187 (2017) 117–127 Table 5 RON and MON values of real fuels from the OITs and the literature. OIT
Nominal/literature
Fuel
RON
MON
S
RON
MON
S
FACE A gasoline FACE F gasoline FACE I gasoline FACE J gasoline Coryton gasoline Haltermann gasoline
82.7 93.4 70.0 73.5 95.0 90.4
82.1 85.9 69.6 70.4 84.9 81.4
0.6 7.5 0.4 3.1 10.1 9.0
83.5 94.4 70.3 71.8 97.5 91.0
83.6 85.8 69.6 68.8 86.6 83.4
0.1 8.6 0.7 3.0 10.9 7.6
screening tool for determining initial RONs and MONs. The k value depends mainly on operating conditions, such as pressure, temperature, and equivalence ratio. This work is the first attempt to determine k values from a combustion device other than an IC engine. Determining these k values is useful for determining what the appropriate conditions are for measuring ignition delay times; this enables a link between the operating conditions of an IQT and IC engines. 3.5. Potential correlation between DCN and RON The ASTM D6890 standard [14] was also used to calculate the DCN of all the fuels used in this study, and the results were used to develop correlations between DCN and RON. Based on the range of the ignition delay times, the DCN was calculated using two equations [14]. For sid in the range of 3.1 to 6.5 ms, the DCN was calculated using Eq. (5a), but outside of this range, DCN is calculated using Eq. (5b). Table 6 Nominal RON and measured DCN of all fuels tested in this study. Fig. 13. Comparison of predicted and nominal values of RON (top) and MON (bottom) of TPRFs and real fuels.
Fuel
Nominal RON
DCN
0.0 50.0 60.0 70.0 80.0 90.0 100.0
53.8 40.1 36.1 31.6 27.7 23.3 17.4
TRF40 TRF45 TRF50 TRF55 TRF60 TRF65 TRF70
53.7 59.7 65.6 71.3 76.9 82.3 87.5
36.7 33.8 31.0 28.6 25.8 22.9 19.8
TPRF TPRF TPRF TPRF TPRF TPRF TPRF TPRF
60-10 60-20 60-30 60-40 70-10 70-20 70-30 70-40
66.2 73.3 79.9 85.9 74.8 80.7 86.1 91.0
32.2 29.1 26.2 22.7 28.9 26.3 23.7 19.9
TPRF TPRF TPRF TPRF
80-10 80-20 80-30 80-40
83.5 88.0 92.2 96.1
25.4 22.8 19.8 16.5
TPRF 90-10 TPRF 90-20 TPRF 90-30
92.0 95.3 98.3
21.1 18.3 15.6
FACE A gasoline FACE F gasoline FACE I gasoline FACE J gasoline Coryton gasoline Haltermann gasoline
83.5 94.4 70.3 71.8 97.5 91.0
26.4 19.1 31.7 29.7 17.3 20.3
PRF PRF PRF PRF PRF PRF PRF
Fig. 14. k values from the KR-IQT at P0 = 21.3 bar, T0 = 770–850 K, and / = 0.7.
tested are represented along with a linear regression model. The best fit for all the fuels (i.e., k values of KR-IQT at P0 = 21.3 bar, T0 = 770–850 K, and / = 0.7) is k = 18.2 (T0/1000) 15.9. The similarity of all k values for all the fuel sets tested at specified charge temperatures indicates the effectiveness of the methodology described here for estimating the OI, the RON, and the MON with an IQT. This emphasizes the prediction capability of the methodology using the KR-IQT, indicating that it is a suitable
0 50 60 70 80 90 100
126
N. Naser et al. / Fuel 187 (2017) 117–127
DCN ¼ 4:46 þ
186:6
ð5aÞ
sid ½ms
DCN ¼ 83:99½ðsid ½ms 1:512Þ
0:658
þ 3:547
ð5bÞ
Correlations between RON and cetane number for PRFs reported in [26] indicate that cetane number correlates with octane number with a cubic regression model. In [1,6] a linear regression model was used on cetane number and octane number of PRFs and TRFs. Table 6 lists the nominal RON and measured DCN of all the fuels used in this study, and Fig. 15 illustrates the excellent correlation between DCN and RON. Quadratic regression for this data set provides Eq. (6). Note that the correlation between DCN and MON is poor as expected because of fuel sensitivity, such that a single DCN cannot simultaneously correlate with both RON and MON:
2 DCN DCN RON ¼ 293 52 þ 114:1 100 100
ð6Þ
Comparisons of these correlations with a cubic regression model by Ryan and Matheaus [26] and a linear regression model by Kalghtagi [1,6] are shown in Fig. 16 along with their R2 values, indicating improved accuracy using the correlation presented in this work.
4. Concluding remarks A new methodology to estimate OI, RON, and MON was developed using ignition delay times from an IQT. The assumption is made that for the same ignition delay time under the same operating conditions, non-PRF fuels should have the same OI as the PRF fuels. Ignition delay times of PRFs and TRFs at a fixed pressure with varying charge temperature were measured in the KR-IQT. By correlating OI with ignition delay time, a functional form given in Eq. (2) was tested and adjusted OITs were determined to estimate RON and MON through the linear extrapolation of OI. The best values for TRON and TMON were 875 K and 930 K, respectively. The methodology was further tested for the TPRFs and the real fuels and the predicted RON/MON values had a sufficient degree of accuracy with the RON/MON values obtained from the standard methods and the literature. The k value, a measure of operating condition from the RON test condition, is independent of the fuels tested and decreased linearly with decreasing charge temperature; this is in line with observations from engine tests [6]. The methodology presented here can be used as a viable initial screening tool for the design, selection, and production of fuel to match the octane appetite of future engines [7]. The methodology may not be limited to IQT measurements, and it may be equally applicable to ignition delay time data from shock tubes and rapid compression machines. Investigations into these future works are underway. Finally, we evidence a correlation between DCN and RON that would be very useful for the combustion community in designing future fuels for future IC engines. Acknowledgments This work was supported by Saudi Aramco under the FUELCOM program and the Clean Combustion Research Center (CCRC) at King Abdullah University of Science and Technology (KAUST). Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.fuel.2016.09.013. References
Fig. 15. Correlation between DCN and RON.
Fig. 16. Comparison of predicted RON from DCN with data found in this study and from correlations available in the literature (values in parentheses indicate R2).
[1] Kalghatgi GT. Fuel/engine interactions. Warrendale, PA: SAE International; 2014. [2] ASTM International. Standard test method for research octane number of spark-ignition engine fuel. ASTM D2699-13b; 2013. [3] ASTM International. Standard test method for motor octane number of sparkignition engine fuel. ASTM D2700-14; 2014. [4] Leppard WR. The chemical origin of fuel octane sensitivity. SAE technical paper; 902137; 1990. [5] Kalghatgi GT. Fuel anti-knock quality – Part I, Engine studies. SAE paper 200101-3584; 2001. [6] Kalghatgi GT. Auto-ignition quality of practical fuels and implications for fuel requirements of future SI and HCCI engines. SAE paper 2005-01-0239; 2005. [7] Kalghatgi GT, Nakata K, Mogi K. Octane appetite studies in direct injection spark ignition (DISI) engines. SAE paper 2005-01-0244; 2005. [8] Kalghatgi GT. Fuel anti-knock quality – Part II, Vehicle studies – How relevant is motor octane number (MON) in modern engines? SAE paper 2001-01-3585; 2001. [9] Mittal V, Heywood JB. The relevance of fuel RON and MON to knock onset in modern SI engines. SAE paper 2008-01-2414; 2008. [10] Bradley D, Head RA. Engine autoignition: The relationship between octane numbers and autoignition delay times. Combust Flame 2006;147:171–84. [11] Perez PL, Boehman AL. Experimental investigation of the autoignition behavior of surrogate gasoline fuels in a constant-volume combustion bomb apparatus and its relevance to HCCI combustion. Energy Fuels 2012;26(10):6106–17. [12] Kalghatgi G, Head R, Chang J, Viollet Y, Babiker H, Amer A. An alternative method based on toluene/n-heptane surrogate fuels for rating the anti-knock quality of practical gasolines. SAE Int J Fuels Lubr 2014;7:663–72. [13] Kalghatgi G, Babiker H, Badra J. A simple method to predict knock using toluene, n-heptane and iso-octane blends (TPRF) as gasoline surrogates. SAE Int J Eng 2015;8(2):505–19.
N. Naser et al. / Fuel 187 (2017) 117–127 [14] ASTM International. Standard test method for determination of ignition delay and derived cetane number DCN of diesel fuel oils by combustion in a constant volume chamber. ASTM D6890; 2008. [15] Yang SY, Naser N, Chung SH, Cha J. Effect of temperature, pressure and equivalence ratio on ignition delay in ignition quality tester (IQT): Diesel, nheptane, and iso-octane fuels under low temperature conditions. SAE Int J Fuels Lubr 2015;8(3):537–48. [16] Bogin G, Dean AM, Ratcliff MA, Luecke J, Zigler BT. Expanding the experimental capabilities of the ignition quality tester for autoigniting fuels. SAE Int J Fuels Lubr 2010;3(1):353–67. [17] Kuti OA, Yang SY, Hourani N, Naser N, Roberts WL, Chung SH, Sarathy SM. A fundamental investigation into the relationship between lubricant composition and fuel ignition quality. Fuel 2015;160:605–13. [18] Rosseel E, Sierens R. The physical and the chemical part of the ignition delay in diesel engines. SAE technical paper 961123; 1996. [19] Bogin Jr GE, Osecky E, Chen JY, Ratcliff MA, Luecke J, Zigler BT, Dean AM. Experiments and computational fluid dynamics modeling analysis of large nalkane ignition kinetics in the ignition quality tester. Energy Fuels 2014;28 (7):4781–94.
127
[20] Yang H, Shuai S, Wang Z, Wang J, Xu H. Performance of straight-run naphtha single-and two-stage combustion modes from low to high load. Int J Engine Res 2013;14(5):469–78. [21] Mittal V, Heywood JB, Green WH. The underlying physics and chemistry behind fuel sensitivity. SAE Int J Fuels Lubr 2010;3(1):256–65. [22] Zheng Z, Badawy T, Henein N, Sattler E. Investigation of physical and chemical delay periods of different fuels in the ignition quality tester. J Eng Gas Turb Power 2013;135:061501. [23] Cleveland WS, Devlin SJ. Locally weighted regression: An approach to regression analysis by local fitting. J Am Stat Assoc 1988;83(403):596–610. [24] Cannella W, Foster M, Gunter G, Leppard WR. FACE gasolines and blends with ethanol: Detailed characterization of physical and chemical properties. CRC report no. AVFL-24; 2014. p. 1–60. [25] Hildingsson L, Kalghatgi G, Tait N, Johansson B, Harrison A. Fuel octane effects in the partially premixed combustion regime in compression ignition engines. SAE paper, 2009-01-2648; 2009. [26] Ryan TW, Matheaus AC. Fuel requirements for HCCI engine operation. SAE paper, 2003-01-1813; 2003.
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Full Length Article
Relating the octane numbers of fuels to ignition delay times measured in an ignition quality tester (IQT) Nimal Naser a,⇑, Seung Yeon Yang a, Gautam Kalghatgi b, Suk Ho Chung a,⇑ a b
King Abdullah University of Science and Technology (KAUST), Clean Combustion Research Center (CCRC), Thuwal, Saudi Arabia Saudi Aramco Research and Development Center, Fuel Technology Research and Development Division, Dhahran, Saudi Arabia
h i g h l i g h t s A methodology for estimating the RON and the MON of fuels using an IQT is presented. Estimated RON and MON are in good agreement with values obtained from standard tests. The DCN of PRFs, TRFs, TPRFs FACE and certification gasolines are reported. Correlation between DCN and RON is provided.
a r t i c l e
i n f o
Article history: Received 5 May 2016 Received in revised form 2 September 2016 Accepted 4 September 2016
Keywords: Octane index Octane numbers Ignition quality tester Autoignition Combustion Gasoline Derived cetane number
a b s t r a c t A methodology for estimating the octane index (OI), the research octane number (RON) and the motor octane number (MON) using ignition delay times from a constant volume combustion chamber with liquid fuel injection is proposed by adopting an ignition quality tester. A baseline data of ignition delay times were determined using an ignition quality tester at a charge pressure of 21.3 bar between 770 and 850 K and an equivalence ratio of 0.7 for various primary reference fuels (PRFs, mixtures of isooctane and n-heptane). Our methodology was developed using ignition delay times for toluene reference fuels (mixtures of toluene and n-heptane). A correlation between the OI and the ignition delay time at the initial charge temperature enabled the OI of non-PRFs to be predicted at specified temperatures. The methodology was validated using ignition delay times for toluene primary reference fuels (ternary mixtures of toluene, iso-octane, and n-heptane), fuels for advanced combustion engines (FACE) gasolines, and certification gasolines. Using this methodology, the RON, the MON, and the octane sensitivity were estimated in agreement with values obtained from standard test methods. A correlation between derived cetane number and RON is also provided. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Engine knock, an abnormal combustion phenomenon that occurs in internal combustion (IC) engines, limits the thermal efficiency of spark ignition (SI) engines [1]. It is caused by autoignition in hot spots at the unburned end gas ahead of the expanding flame front and depends on pressure and temperature evolution in the end gas as well as on the anti-knock quality of the fuel. The autoignition or anti-knock quality of a fuel is measured by the research octane number (RON) and the motor octane number (MON), which are determined using the standard test methods prescribed by ASTM D2699 and D2700 [2,3], respectively, with a ⇑ Corresponding authors. E-mail addresses: [email protected] (N. Naser), [email protected]. sa (S.H. Chung). http://dx.doi.org/10.1016/j.fuel.2016.09.013 0016-2361/Ó 2016 Elsevier Ltd. All rights reserved.
cooperative fuel research (CFR) engine. In these tests, the knock behavior of a fuel is compared to that of primary reference fuels (PRFs), mixtures of iso-octane and n-heptane, with known isooctane volume percentages. The RON (or the MON) of the test fuel can then be related to that of the PRF that matches its knock behavior. There are, however, differences in the chemical composition and autoignition chemistry of practical fuels and the PRFs that are used to define the octane scale. In addition, the same gasoline may match different PRFs at different operating conditions. For all practical fuels, the RON is higher than the MON because the temperature of an unburned gas is higher at a given pressure under MON than under RON conditions. This is due to the more severe MON testing conditions, which are performed at an intake mixture temperature of 422 K and an engine speed of 900 rpm as compared with RON testing conditions, which are performed at 325 K and
118
N. Naser et al. / Fuel 187 (2017) 117–127
600 rpm. As a result, in MON testing, the autoignition chemistry of PRFs is dominated by negative temperature coefficient (NTC) behavior [4]. Practical fuels, at similar conditions, however, do not exhibit NTC behavior and therefore demonstrate the antiknock characteristics of a lower-octane-quality PRF than what was used in RON testing: RON values are higher than MON values [4]. The anti-knock quality of a fuel can best be described by the octane index (OI) [1,5–8], which is defined as
OI ¼ ð1 kÞRON þ kMON
ð1aÞ
OI ¼ RON kS
ð1bÞ
Here, the weighting factor k is a constant depending on pressure/ temperature histories with time in an end gas [1] and engine speed, and is weakly dependent on mixture strength [1,9]; S is the octane sensitivity defined as (RONMON); and octane index (OI) is the volume percent of iso-octane in the PRF that matches the knock behavior of the fuel at any given condition. By definition, k = 1 under MON testing conditions and k = 0 under RON testing conditions. If the temperature of an unburned gas for a specified pressure decreases, the k value decreases [1,6]. Sensitivity (S) can be regarded as a measure of the difference in autoignition chemistry of a fuel at the two operating conditions, while k is a measure of how different the operating conditions are between the two tests. The temperature of the unburned bulk gas at 15 bar designated as Tcomp15 was introduced in [6], and this information was used to establish a given operating condition with respect to the RON test condition. Empirically, it has been demonstrated that the k value decreases as Tcomp15 decreases [1,6]. SI engine design has developed to enable higher pressure for a given temperature to achieve better thermal efficiency and power density. So historically, Tcomp15 has decreased from the value representative of the MON test such that the importance of MON in describing fuel anti-knock quality has decreased [1,9]. In fact, Tcomp15 in modern engines is even lower than in the RON test condition and k can be negative, and thus for a given RON, a fuel with lower MON has higher OI and is more resistant to knock. Further efforts to improve SI engine efficiency by downsizing and turbocharging are driving the trend towards more negative k values, while old engines and CFR engines have positive k values [9]. Attempts have been made to interpret the RON and the MON of practical fuels in terms of more fundamental measures of autoignition, such as ignition delay time. Bradley et al. [10] showed that ignition delay times of non-PRFs can be deduced reasonably from the OIs measured in a variety of engine tests using the data available on ignition delay times for PRFs. Analogous to the k weighting factor relating OI to RON and MON, they defined the j factor relating the ignition delay times of non-PRFs to those for PRFs under the same conditions. Perez and Boehman [11] measured the ignition delay times in an ignition quality tester (IQT) for 21 surrogate gasoline fuels to find a poor correlation between the measured ignition delay time with the RON and the MON. Kalghatgi et al. [12] suggested that fuels should be rated on a toluene/n-heptane scale rather than using PRFs because with toluene/n-heptane blends, toluene reference fuels (TRF), the sensitivity will be high and comparable to that of practical fuels. Using the current RON test, a fuel could be assigned a toluene number by the volume percent of toluene in the TRF blend matching the fuel behavior in the RON test. However, it is possible to find a ternary mixture of toluene, iso-octane, and n-heptane (toluene/PRF or toluene primary reference fuel (TPRF) blends) that matches both the RON and the MON of a fuel. Kalghatgi et al. [13] developed empirical correlations to predict the RON, the MON, and octane sensitivity for TPRF blends based on experimental RON/MON measurements of such mixtures.
Standard RON and MON tests are time consuming and require strict control of various parameters involved during engine operation. The methodology presented in this paper, which can be a convenient initial screening tool to determine RON and MON, estimates the RON and the MON from a combustion device other than the CFR engine. This methodology was developed using a constant volume combustion chamber with liquid fuel spray injection, which is capable of operating at various temperatures and pressures. For wide applicability of the developed methodology, a constant volume combustion chamber of a standard IQT (Advanced Engine Technology Ltd.) system was adopted, which is used to determine the derived cetane number (DCN) of diesel fuels based on ASTM D6890 [14] at the charge temperature and pressure of 818 ± 30 K and 21.37 ± 0.07 bar, respectively. The ignition delay time, sid, was measured as the time between the start of fuel injection and the start of significant heat release; an average of 32 firings were taken. The DCN is related to the ignition delay time in [ms] by the formula DCN = 4.46 + (186.6/sid) for ignition delay times between 3.1 and 6.5 ms, which correspond to DCNs between 64 and 33. The KAUST research IQT (KR-IQT) was modified from the standard IQT, which typically has a fixed volume of injected fuel. Since equivalence ratio is an important parameter affecting ignition delay times, we have made modifications to testing methods adopted in this study as described in [15] to include this ratio (details in the next section). Ignition delay times for fuels in the diesel autoignition range are typically short and thus contain both physical (evaporation and mixing) and chemical (reactive) time contributions. For fuels in the gasoline autoignition range, RON > 60 and DCN < 30 [6], with relatively long ignition delay times, the physical delay may be much shorter than the chemical delay, which is one motivation for using the KR-IQT for determining OI. Ignition delay times for the fuel mixtures of PRFs were measured first to provide a baseline data set. Ignition delay time data for octane sensitive TRFs were then used to develop a methodology. Fuel mixtures of toluene primary reference fuels (TPRFs), fuels for advanced combustion engines (FACEs), and certification gasolines were used to validate the methodology.
2. Experiment A schematic of the experimental setup is shown in Fig. 1. The apparatus comprises a KR-IQT, a gas supply system, and several data acquisition setups. The KR-IQT is equipped with a constant volume combustion chamber at 0.21 L, filled with an oxidizing medium that can be pressurized up to a maximum of 21.6 bar and heated with an electric heater to temperatures in the range of 670–860 K. A liquid fuel spray is injected into a pressurized and heated oxidizing environment. A pneumatically driven mechanical pump delivers fuel stored in a fuel reservoir to the injector with a single-hole S-type inward opening pintle nozzle [16]. The temperature of the nozzle tip during operation is around 325 K. The injected fuel mass can be regulated by adopting a pneumatically driven variable displacement pump (VDP), which varies displacement of the pump, thereby the volume of fuel injected. This modification was provided by the manufacturer. It uses turnbuckle bolts, which when rotated, change the displacement between the front body and rear body of VDP using a chain system as was shown in Fig. 3 in [15]. The VDP has a limited displacement range, therefore, for a specified charge pressure, there is a limited range of equivalence ratio available depending on the density and oxygen requirement of the fuel. More details on mass calibration for the VDP have been explained in detail in [15,17]. The fuel is injected into the combustion chamber at a pressure of 225 bar [16]. A piezo-electric pressure transducer mounted
N. Naser et al. / Fuel 187 (2017) 117–127
119
Fig. 1. Schematic of the experimental setup.
along the same axis as that of the fuel injector acquires a chamber pressure signal, which is collected by a data acquisition system along with the needle lift values from a needle lift sensor placed in front of the injector. These two signals form the backbone of measuring ignition delay times of test fuels. A total ignition delay time, sid , includes delays due to both physical and chemical processes during ignition. Physical ignition delay time describes the time required for the injected liquid spray to atomize into fine droplets, heat, evaporate, and mix with oxidizer to form a fuel-oxidizer mixture that can ignite. Chemical ignition delay time describes the time required for fuel molecules to breakdown into smaller fragments and react with oxidizer molecules to form sufficient chain branching radicals that ignite the mixture [18]. The ignition delay times presented throughout this paper are total ignition delay times. As mentioned previously, ignition delay time is relatively much longer for gasoline fuels than for diesel fuels such that the contribution from a physical process is expected to be relatively small. The total ignition delay time is defined as the time difference between the start of injection (SoInj) and the start of ignition (SoIgn), as indicated in Fig. 2. The SoInj was set as the point where the needle lift attains its maximum. The SoIgn is determined differently from the conventional IQT system software and modified definitions are used [15]. In the conventional IQT software, SoIgn is determined as the point where the difference in chamber pressure P and the initial charge pressure P0 of DP exceeds a preset value of 1.38 bar. This definition is suitable for diesel-like fuels, where distinctive low-temperature heat release is not conspicuous due to shorter ignition time scales. This definition is, however, unsuitable for fuels that exhibit low-temperature heat-release behavior because the point of ignition is determined to be much earlier than the actual point of ignition, as demonstrated in Fig. 2. Here, the chamber pressure is shown as a function of time t for iso-octane fuel, and the conventional IQT definition, represented by the vertical dotted line marked as SoIgn (IQT), is shown to be unable to correctly represent the rapid pressure rise near
Fig. 2. Typical change in pressure with time of iso-octane ignition at P0 = 21.3 bar, T0 = 720 K, and / = 0.7, indicating various definitions of SoIgn.
t = 37 ms. Note that this delay time is much longer than the previously mentioned typical range of the DCN test of 3.1–6.5 ms. To compensate for the effect of the low-temperature heat-releaseinduced pressure rise, Bogin et al. [19] used a preset value of 2.76 bar with this same criteria, represented by the vertical line as SoIgn (Bogin et al.), but they failed to correctly predict rapid pressure rises. Instead, the gradient method proposed in [15] is used to accurately determine the start of ignition while allowing for the effects of low-temperature heat release. This method determines the point of ignition as the intersection of two line tangents (LT) in a pressure–time (P-t) diagram. The first line tangent (dP/dt) (LT 1) is determined at the point where the chamber regains its initial pressure. As shown in Fig. 2, the pressure initially decreases due to evaporative cooling of the liquid fuel spray droplets, but as the chemical reaction proceeds, the pressure returns to the
120
N. Naser et al. / Fuel 187 (2017) 117–127
Fig. 3. Ignition delay times of PRFs.
initial level. The second line tangent (LT 2) is selected as the maximum slope of the pressure rise during ignition. The point where these two line tangents intersect is designated as the SoIgn (gradient method). This method successfully captured the timing of rapid pressure rise. All fuel samples were measured at P0 = 21.3 bar with an equivalence ratio of / = 0.7, a pressure required by the DCN standard ASTM D6890 [14]. Rather than fixing the volume of the injected fuel as in determining DCN, we fixed the equivalence ratio. This approach was successful because the VDP accurately evaluated the equivalence ratio window for each fuel despite differences in fuel densities and stoichiometric air-fuel ratios. Selection of the equivalence ratio at / = 0.7 was based on the ability of this value to represent a gasoline compression ignition condition for advanced engines with the lean-burn strategy [20]. Since many gasoline engines are operated at stoichiometric conditions, cases using an equivalence ratio of / = 1.0 were also tested for certain fuels and the results are presented in the Supplementary Material. We selected an initial pressure of P0 = 15 bar, motivated by the definition of Tcomp15 in [6], relating the temperature of mixture at 15 bar during engine experiment to RON and MON values. This pressure is therefore likely ideal for correlating IQT conditions to IC engine conditions during stoichiometric engine operations. As mentioned previously, this pressure condition satisfies the limitations on the displacement range of VDP at / = 1.0. The charge temperature is varied from 770 to 850 K, which, for most modern engines, is a temperature range that corresponds to compression temperatures [21]. The zero air oxidizer was used for all tests (air containing trace amounts of total hydrocarbons <0.1 ppm) with an O2 concentration of 20.9 ± 1% balanced with nitrogen. All experiments are conducted following the standard procedure specified in ASTM D6890. For each test condition, 15 pre-injections are carried out to establish a stable, uniform, steady environment for the 32 main injections that follow. The data of these main injections were used for processing. A MATLAB code was developed to accommodate the new definition of SoIgn we described. The raw chamber pressure signal is smoothed or filtered to eliminate fluctuations in the signal due to resonance and noise, as can be seen in Fig. 2. Although severe pressure oscillations were observed mainly after the rapid pressure rise in Fig. 2, there were also non-negligible pressure fluctuations during the induction period. Zheng et al. [22] employed an eleven-point moving average and a finite impulse response filter with phase correction to remove some important frequency com-
ponents; however, these adjustments failed to satisfactorily eliminate the frequency of some of the components that had to be removed. Thus, the data processing code comprised a local regression using weighted linear least squares and a second-degree polynomial model (locally weighted scatterplot smoothing, LOESS) [23]. Details on data processing are available in [15]. Here, ignition delay time was determined for four sets of fuel samples, as specified in Tables 1 and 2. The first set comprised PRFs, whereby PRF X indicates a mixture of an X volume percentage of iso-octane and the remaining (100 X) volume percentage of n-heptane to provide a baseline data set. The second set comprised TRFs, whereby TRF X contains an X volume percentage of toluene. The third set comprised TPRFs, where TPRF X-Y indicates a mixture of PRF X with a Y volume percentage of toluene, and the fourth set comprised FACE gasolines [24] and the certification gasoline fuels listed in Table 2 (collectively called real fuels). The RON and the MON of TRFs and TPRFs were obtained from the standard ASTM tests conducted in [12,13], and when experimental octane numbers were not available, they were calculated with empirical correlations developed by Kalghatgi et al. [13]. The RON and the MON of the FACE and the certification gasolines were obtained from the certificate of analysis provided by the fuel suppliers. All fuels selected for this study had RON/MON values greater than 40 because the standard method for determining octane number from a CFR engine have a digital counter setting for octane numbers above 40 and hence can only determine the octane number for fuels that have RON/MON values above 40.
3. Methodology for estimating RON, MON, and OI 3.1. Baseline data for PRFs and TRFs We sought to answer whether ignition delay times obtained from IQT experiments could be used to estimate RON, MON, and OI. Thus, we made the assumption that fuels with the same octane index operating under the same pressure/temperature history have the same ignition delay time [12]. We base this basic premise of the methodology on understanding the variation of ignition delay times of a non-PRF with respect to PRFs at different temperatures. Depending on the sensitivity of a fuel, a non-PRF can match different PRFs at different temperatures if it has considerable sensitivity. In such a case, best fit temperatures from OI data will be selected, which can estimate RON/MON values. Note that a nonPRF with zero sensitivity can match a particular PRF at all temperatures. From this assumption, we aimed to find the OI of non-PRFs at various temperatures while maintaining all other parameters constant, including charge pressure, equivalence ratio, and injection pressure. First, the ignition delay times of PRFs were measured using the KR-IQT to determine the relationship between OI and ignition delay times at various initial temperatures at 21.3 bar and / = 0.7. The ignition delay times of PRFs against the reciprocal of charge temperature 1000/T0 are shown in Fig. 3. Results showed that the logarithm of the total ignition delay increases as the charge temperature (T0) decreases in a reasonably linear manner, indicating an Arrhenius-type temperature dependence; ignition delay time also increases with PRF X. The ignition delay times of non-PRFs are also tested at the same initial conditions and the results for the TRF fuels are shown in Fig. 4 and the PRF data from Fig. 3 are also shown with dotted lines. We observed that TRFs show a similar trend to that for PRFs: the logarithm of ignition delay time increases linearly with decreasing charge temperature. Note that a fuel that has zero or little sensitivity has an ignition delay curve that is nearly parallel to the corresponding PRF, as will be shown later when we validate our
121
N. Naser et al. / Fuel 187 (2017) 117–127 Table 1 Fuel matrix of PRFs, TRFs, and TPRFs. Fuel
*
PRF
Toluene vol.%
Vol.%
MON
RON/MON Refs.
Toluene
iso-octane
n-heptane
PRF PRF PRF PRF PRF PRF PRF
0 50 60 70 80 90 100
0 50 60 70 80 90 100
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 50 60 70 80 90 100
100 50 40 30 20 10 0
0.0 50.0 60.0 70.0 80.0 90.0 100.0
0.0 50.0 60.0 70.0 80.0 90.0 100.0
[2,3] [2,3] [2,3] [2,3] [2,3] [2,3] [2,3]
TRF TRF TRF TRF TRF TRF TRF
40 45 50 55 60 65 70
0 0 0 0 0 0 0
40 45 50 55 60 65 70
40 45 50 55 60 65 70
0 0 0 0 0 0 0
60 55 50 45 40 35 30
53.7 59.7 65.6 71.3 77.9 84.5 89.3
47.1 52.4 57.6 62.6 68.5 74.5 78.2
BR* [13] BR [13] [12] BR [13] [12] [12] [12]
TPRF TPRF TPRF TPRF
60-10 60-20 60-30 60-40
60 60 60 60
10 20 30 40
10 20 30 40
54 48 42 36
36 32 28 24
66.0 73.6 79.0 86.2
64.4 70.0 74.0 79.6
[13] [13] [13] [13]
TPRF TPRF TPRF TPRF
70-10 70-20 70-30 70-40
70 70 70 70
10 20 30 40
10 20 30 40
63 56 49 42
27 24 21 18
74.8 80.7 86.1 91.0
72.8 76.9 80.6 84.0
BR BR BR BR
TPRF TPRF TPRF TPRF
80-10 80-20 80-30 80-40
80 80 80 80
10 20 30 40
10 20 30 40
72 64 56 48
18 16 14 12
84.5 89.1 92.8 96.7
82.0 85.6 86.9 88.7
[13] [13] [13] [13]
TPRF 90-10 TPRF 90-20 TPRF 90-30
90 90 90
10 20 30
10 20 30
81 72 63
9 8 7
92.0 95.3 98.3
90.0 91.4 92.7
BR [13] BR [13] BR [13]
[13] [13] [13] [13]
BR indicates blending rule.
Table 2 Fuel matrix of FACE and certification gasolines.
*
RON
Fuel
RON
MON
Ref.
FACE A FACE F FACE I FACE J Coryton gasoline Haltermann gasoline
83.5 94.4 70.3 71.8 97.5 91.0
83.6 85.8 69.6 68.8 86.6 83.4
COA* COA COA COA COA COA
COA indicates certificate of analysis.
methodology with real fuels having zero or low sensitivity. Note that although the slopes of all TRFs appear to be similar, this is an artifact created by plotting them on a logarithmic scale. To correct for this artifact, we plotted this data on a linear scale, which highlights the difference in their slopes (Fig. S6 in Supplementary Material). Similarly, a linear scale shows the differences in the slopes of PRFs are (Fig. S6 in Supplementary Material). As the sensitivity of the fuel increases, the ignition delay time curves intersect with the PRF X curve. In other words, the difference in slope between ignition delay time curve of an unknown fuel and the adjacent PRF curves is related to the fuel’s octane sensitivity. The TRF data exhibits this behavior because toluene is a component that influences sensitivity to the resulting mixture. 3.2. Correlation with OI Known OI values for PRFs are plotted as a function of ignition delay times in Fig. 5. Note that the RON and the MON values of PRFs are the same by definition and their sensitivities are 0; as a result OI = RON = MON. Thus, for a specified PRF, OI does not vary at different charge temperatures. For a specified temperature, there is a gradual reduction in the slope [d(OI)/dsid] as ignition delay time increases. For a fixed charge temperature, OI can be related to ignition delay time by the following formula:
OI ¼ a½ln sid þ b: n
Fig. 4. Ignition delay times of TRFs.
ð2Þ
As will be shown later in this section, data representation in the functional form of Eq. (2) has a linear relationship with temperature in the OI against the [lnsid]n plot. The exponent n was selected by an ordinary least-squares approach by creating a loop in a range of 10 6 n 6 10 with a step
122
N. Naser et al. / Fuel 187 (2017) 117–127
Fig. 5. OI of PRFs at several temperatures.
Fig. 7. OI of TRFs.
of 0.1 to obtain maximum coefficients of determination. The maximum coefficient of determination R2 is obtained when n = 2.4. Fig. 6 shows that the expected linear relationship between OI n and ½lnsid assumed in Eq. (2) does indeed hold, evidenced by the linear isotherms. The best fit parameters of a and b are both dependent on temperature, and they have the quadratic variation as given by the following equations (comparison with other regression models are shown in Fig. S7 in Supplementary Material):
for a given T0 (isotherm) and the measured value ignition delay time (vertical line from the horizontal axis). Fig. 7 shows the estimates of OIs of TRFs at several temperatures determined in this manner. The OI values of all TRFs decrease with increasing temperature. The OIs of TRFs with higher toluene content are observed to
The OI of a TRF can be determined using the measured ignition delay time with the initial charge temperature. This assumption is based on the premise that for the same ignition delay time under the same operating conditions, non-PRFs and PRFs have the same OI. On an OI versus [ln sid]2.4 plot with the same isotherms as those in Fig. 6, OI for a TRF can be determined from the cross point
decrease more rapidly with ½lnsid for the same temperature change, which is an indication of the octane sensitivity of the fuels. Alternatively, the OI of TRFs varies with ignition delay time as temperature varies. The slope for the specified charge temperature in Fig. 7 is a qualitative indication of the k value given in Eq. (1). As temperature increases, the absolute value of the slope of the isothermal line decreases, indicating a smaller change in OI with change in ignition delay. At relatively lower temperatures, the slope increases where the change in OI with ignition delay is higher. Note that the isotherms intersect for OI values greater than about 100, but this cross-over of isotherms does not pose a problem in determining estimated RON and MON values, as we will demonstrate next. If the OI from Fig. 7 is plotted against the charge temperature, there is reasonably linear variation with the temperature for each TRF, as shown in Fig. 8. The decrease in OI or autoignition quality of
Fig. 6. OI versus [ln sid]2.4 of PRFs exhibiting linear iso-T0 lines (values in parentheses indicate R2).
Fig. 8. OI versus charge temperature demonstrating the extrapolation method to estimate RON and MON from IQT data for TRFs.
2 T0 T0 a ¼ 39862 þ 70652 31456 1000 1000
ð3aÞ
2 T0 T0 2216 þ 1190 b ¼ 1121 1000 1000
ð3bÞ
2:4
123
N. Naser et al. / Fuel 187 (2017) 117–127
values from the literature designated as RONnominal and MONnominal in Fig. 9. The nominal values were obtained using the standard tests [2,3] obtained in [12,13] and by the correlations in [13], which are also listed in Table 3 together with the OIT values. These results demonstrate that the estimated OIT values are in good agreement with the nominal values as shown by the correlation coefficients of R2 = 0.965 and 0.878 for RON and MON, respectively.
the fuel with increasing charge temperature is evident. The method of determining the RON and the MON from these data is described as follows. As can be observed in Fig. 8, the OI values (solid symbols) of all the fuels are higher than their corresponding nominal RON and MON values given in Table 1. A linear extrapolation of the OI curve for each fuel (solid lines) to temperatures beyond the operating temperature range of KR-IQT indicates that OI values decrease at higher temperatures. A linear regression fit of each OI curve and a substitution of RON and MON values provides the temperatures at which the estimated OI values match nominal RON and MON values. Note that by adjusting the temperature for each fuel, the temperature at which the OI matches the RON/MON value exactly can be identified, although it could be slightly different for different fuels; however, temperatures are always similar, leading to small differences in estimated RON and MON values if this common fixed temperature is used. In line with the above discussion, two common fixed octane index temperatures (OIT) that provide best estimates of RON and MON values are identified at 875 K and 930 K for RON and MON, respectively. For an octane sensitive fuel such as a TRF, the OI decreases as temperature increases at a fixed pressure (i.e., it becomes more prone to autoignition and knock). In Fig. 8, the value of k (see Eq. (1a)) is 0 and 1 at 875 K and 930 K, respectively, and the slope of the line for each TRF depends on its octane sensitivity, S, which increases as the toluene concentration in the TRF increases. The estimated RON and MON values obtained with the OITs designated as RONOIT and MONOIT are compared with the nominal
3.3. Validation with TPRFs and real fuels To validate this method further, ignition delay times were measured for TPRFs and real fuels at four specified initial temperatures from the IQT experiments. Of the FACE gasolines, FACE A has the lowest octane number with very small sensitivity (S = 0.1) and FACE F has the highest octane number. FACEs I and J have a RON of about 70, and FACE I has a lower sensitivity. Both can be considered suitable fuels for gasoline compression ignition engines [25]. Figs. 10 and 11 show the ignition delay times with the inverse of the initial temperature for TPRFs and real fuels, respectively. To determine their OIs, the same methodology was applied to TPRFs and real fuels as was applied to TRFs. For the sake of clarity only selected TPRFs are shown in Figs. 10 and 12(b). Fig. 10 shows that as the volume percentage of toluene increases, the slopes of TPRFs gradually increase from the slopes of PRFs; for example, TPRF-60-10 (10%vol toluene) has an ignition delay time curve nearly identical to PRF 70, while TPRF-70-40 (40% vol toluene) has larger slope than PRF 90, indicating higher octane sensitivity. Similarly, in Fig. 11, low-octane sensitivity FACE I has a nearly identical slope to that of PRF 70. Meanwhile high-octane sensitivity
Fig. 9. Comparison of predicted and nominal values of RON (top) and MON (bottom) of TRFs.
Fig. 10. Ignition delay times of TPRFs.
Table 3 RON and MON values of TRFs from the OITs and the literature. OIT Fuel TRF TRF TRF TRF TRF TRF TRF
40 45 50 55 60 65 70
Nominal/literature
RON
MON
S
RON
MON
S
54.3 61.6 67.6 73.4 79.1 85.0 90.1
48.0 55.2 61.1 66.4 71.3 76.8 80.6
6.3 6.4 6.5 7.0 7.8 8.2 9.5
53.7 59.7 65.6 71.3 77.9 84.5 89.3
47.1 52.4 57.6 62.6 68.5 74.5 78.2
6.6 7.3 8.0 8.7 9.4 10 11.1
124
N. Naser et al. / Fuel 187 (2017) 117–127
temperature plots in Fig. 12. The data are shown in Tables 4 and 5 with the nominal values from [13] from the CFR tests and the data from the certificates of analysis for the TPRFs and real fuels, respectively. The near horizontal lines on the plot evidence the low sensitivity of both FACE A and I gasolines. The estimated RON and MON values match well with the nominal values, and this result is further exhibited in Fig. 13. The correlation coefficients are acceptable for the initial screening of TPRF and real fuels for the RON and the MON prediction from an IQT experiment.
3.4. Weighting factor k From the RON and the MON values obtained using the methodology described here, it is possible to determine k values given in Eq. (1) as a function of T0,
k¼ Fig. 11. Ignition delay times of real fuels.
Coryton gasoline has a larger slope than PRF 100, indicating a greater change in OI with temperature variation. This same methodology was used to estimate the RON and the MON for the TPRFs and the real fuels using the OI versus charge
RON OI RON MON
ð4Þ
The values of k obtained as a function of charge temperature are shown in Fig. 14 for all the fuels tested. The results show that k values are reasonably independent of the TRFs, the TPRFs, and the real fuels tested and that they depend on the physical operating condition of charge temperature, as per the definition of OI. The respective mean values, kmean, for the TRFs, the TPRFs, and the real fuels
(a)
(b)
Fig. 12. Predicted RONs and MONs of (a) TPRFs and (b) real fuels.
Table 4 RON and MON values of TPRFs from the OITs and the literature. OIT Fuel
Nominal/literature
RON
MON
S
RON
MON
S
TPRF TPRF TPRF TPRF
60-10 60-20 60-30 60-40
67.8 75.5 81.5 88.2
65.9 73.2 77.9 83.4
1.8 2.3 3.6 4.8
66.0 73.6 79.0 86.2
64.4 70.0 74.0 79.6
1.6 3.6 5.0 6.6
TPRF TPRF TPRF TPRF
70-10 70-20 70-30 70-40
76.5 82.4 87.5 92.2
75.3 80.7 85.0 85.1
1.2 1.7 2.5 7.0
74.8 80.7 86.1 91.0
72.8 76.9 80.6 84.0
2.0 3.8 5.5 7.0
TPRF TPRF TPRF TPRF
80-10 80-20 80-30 80-40
84.1 88.8 92.8 97.1
82.0 85.8 86.4 88.5
2.1 3.1 6.4 8.6
84.5 89.1 92.8 96.7
82.0 85.6 86.9 88.7
2.5 3.5 5.9 8.0
TPRF 90-10 TPRF 90-20 TPRF 90-30
92.7 95.6 98.5
90.4 89.7 89.8
2.3 5.9 8.7
92.0 95.3 98.3
90.0 91.4 92.7
2.0 3.9 5.6
125
N. Naser et al. / Fuel 187 (2017) 117–127 Table 5 RON and MON values of real fuels from the OITs and the literature. OIT
Nominal/literature
Fuel
RON
MON
S
RON
MON
S
FACE A gasoline FACE F gasoline FACE I gasoline FACE J gasoline Coryton gasoline Haltermann gasoline
82.7 93.4 70.0 73.5 95.0 90.4
82.1 85.9 69.6 70.4 84.9 81.4
0.6 7.5 0.4 3.1 10.1 9.0
83.5 94.4 70.3 71.8 97.5 91.0
83.6 85.8 69.6 68.8 86.6 83.4
0.1 8.6 0.7 3.0 10.9 7.6
screening tool for determining initial RONs and MONs. The k value depends mainly on operating conditions, such as pressure, temperature, and equivalence ratio. This work is the first attempt to determine k values from a combustion device other than an IC engine. Determining these k values is useful for determining what the appropriate conditions are for measuring ignition delay times; this enables a link between the operating conditions of an IQT and IC engines. 3.5. Potential correlation between DCN and RON The ASTM D6890 standard [14] was also used to calculate the DCN of all the fuels used in this study, and the results were used to develop correlations between DCN and RON. Based on the range of the ignition delay times, the DCN was calculated using two equations [14]. For sid in the range of 3.1 to 6.5 ms, the DCN was calculated using Eq. (5a), but outside of this range, DCN is calculated using Eq. (5b). Table 6 Nominal RON and measured DCN of all fuels tested in this study. Fig. 13. Comparison of predicted and nominal values of RON (top) and MON (bottom) of TPRFs and real fuels.
Fuel
Nominal RON
DCN
0.0 50.0 60.0 70.0 80.0 90.0 100.0
53.8 40.1 36.1 31.6 27.7 23.3 17.4
TRF40 TRF45 TRF50 TRF55 TRF60 TRF65 TRF70
53.7 59.7 65.6 71.3 76.9 82.3 87.5
36.7 33.8 31.0 28.6 25.8 22.9 19.8
TPRF TPRF TPRF TPRF TPRF TPRF TPRF TPRF
60-10 60-20 60-30 60-40 70-10 70-20 70-30 70-40
66.2 73.3 79.9 85.9 74.8 80.7 86.1 91.0
32.2 29.1 26.2 22.7 28.9 26.3 23.7 19.9
TPRF TPRF TPRF TPRF
80-10 80-20 80-30 80-40
83.5 88.0 92.2 96.1
25.4 22.8 19.8 16.5
TPRF 90-10 TPRF 90-20 TPRF 90-30
92.0 95.3 98.3
21.1 18.3 15.6
FACE A gasoline FACE F gasoline FACE I gasoline FACE J gasoline Coryton gasoline Haltermann gasoline
83.5 94.4 70.3 71.8 97.5 91.0
26.4 19.1 31.7 29.7 17.3 20.3
PRF PRF PRF PRF PRF PRF PRF
Fig. 14. k values from the KR-IQT at P0 = 21.3 bar, T0 = 770–850 K, and / = 0.7.
tested are represented along with a linear regression model. The best fit for all the fuels (i.e., k values of KR-IQT at P0 = 21.3 bar, T0 = 770–850 K, and / = 0.7) is k = 18.2 (T0/1000) 15.9. The similarity of all k values for all the fuel sets tested at specified charge temperatures indicates the effectiveness of the methodology described here for estimating the OI, the RON, and the MON with an IQT. This emphasizes the prediction capability of the methodology using the KR-IQT, indicating that it is a suitable
0 50 60 70 80 90 100
126
N. Naser et al. / Fuel 187 (2017) 117–127
DCN ¼ 4:46 þ
186:6
ð5aÞ
sid ½ms
DCN ¼ 83:99½ðsid ½ms 1:512Þ
0:658
þ 3:547
ð5bÞ
Correlations between RON and cetane number for PRFs reported in [26] indicate that cetane number correlates with octane number with a cubic regression model. In [1,6] a linear regression model was used on cetane number and octane number of PRFs and TRFs. Table 6 lists the nominal RON and measured DCN of all the fuels used in this study, and Fig. 15 illustrates the excellent correlation between DCN and RON. Quadratic regression for this data set provides Eq. (6). Note that the correlation between DCN and MON is poor as expected because of fuel sensitivity, such that a single DCN cannot simultaneously correlate with both RON and MON:
2 DCN DCN RON ¼ 293 52 þ 114:1 100 100
ð6Þ
Comparisons of these correlations with a cubic regression model by Ryan and Matheaus [26] and a linear regression model by Kalghtagi [1,6] are shown in Fig. 16 along with their R2 values, indicating improved accuracy using the correlation presented in this work.
4. Concluding remarks A new methodology to estimate OI, RON, and MON was developed using ignition delay times from an IQT. The assumption is made that for the same ignition delay time under the same operating conditions, non-PRF fuels should have the same OI as the PRF fuels. Ignition delay times of PRFs and TRFs at a fixed pressure with varying charge temperature were measured in the KR-IQT. By correlating OI with ignition delay time, a functional form given in Eq. (2) was tested and adjusted OITs were determined to estimate RON and MON through the linear extrapolation of OI. The best values for TRON and TMON were 875 K and 930 K, respectively. The methodology was further tested for the TPRFs and the real fuels and the predicted RON/MON values had a sufficient degree of accuracy with the RON/MON values obtained from the standard methods and the literature. The k value, a measure of operating condition from the RON test condition, is independent of the fuels tested and decreased linearly with decreasing charge temperature; this is in line with observations from engine tests [6]. The methodology presented here can be used as a viable initial screening tool for the design, selection, and production of fuel to match the octane appetite of future engines [7]. The methodology may not be limited to IQT measurements, and it may be equally applicable to ignition delay time data from shock tubes and rapid compression machines. Investigations into these future works are underway. Finally, we evidence a correlation between DCN and RON that would be very useful for the combustion community in designing future fuels for future IC engines. Acknowledgments This work was supported by Saudi Aramco under the FUELCOM program and the Clean Combustion Research Center (CCRC) at King Abdullah University of Science and Technology (KAUST). Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.fuel.2016.09.013. References
Fig. 15. Correlation between DCN and RON.
Fig. 16. Comparison of predicted RON from DCN with data found in this study and from correlations available in the literature (values in parentheses indicate R2).
[1] Kalghatgi GT. Fuel/engine interactions. Warrendale, PA: SAE International; 2014. [2] ASTM International. Standard test method for research octane number of spark-ignition engine fuel. ASTM D2699-13b; 2013. [3] ASTM International. Standard test method for motor octane number of sparkignition engine fuel. ASTM D2700-14; 2014. [4] Leppard WR. The chemical origin of fuel octane sensitivity. SAE technical paper; 902137; 1990. [5] Kalghatgi GT. Fuel anti-knock quality – Part I, Engine studies. SAE paper 200101-3584; 2001. [6] Kalghatgi GT. Auto-ignition quality of practical fuels and implications for fuel requirements of future SI and HCCI engines. SAE paper 2005-01-0239; 2005. [7] Kalghatgi GT, Nakata K, Mogi K. Octane appetite studies in direct injection spark ignition (DISI) engines. SAE paper 2005-01-0244; 2005. [8] Kalghatgi GT. Fuel anti-knock quality – Part II, Vehicle studies – How relevant is motor octane number (MON) in modern engines? SAE paper 2001-01-3585; 2001. [9] Mittal V, Heywood JB. The relevance of fuel RON and MON to knock onset in modern SI engines. SAE paper 2008-01-2414; 2008. [10] Bradley D, Head RA. Engine autoignition: The relationship between octane numbers and autoignition delay times. Combust Flame 2006;147:171–84. [11] Perez PL, Boehman AL. Experimental investigation of the autoignition behavior of surrogate gasoline fuels in a constant-volume combustion bomb apparatus and its relevance to HCCI combustion. Energy Fuels 2012;26(10):6106–17. [12] Kalghatgi G, Head R, Chang J, Viollet Y, Babiker H, Amer A. An alternative method based on toluene/n-heptane surrogate fuels for rating the anti-knock quality of practical gasolines. SAE Int J Fuels Lubr 2014;7:663–72. [13] Kalghatgi G, Babiker H, Badra J. A simple method to predict knock using toluene, n-heptane and iso-octane blends (TPRF) as gasoline surrogates. SAE Int J Eng 2015;8(2):505–19.
N. Naser et al. / Fuel 187 (2017) 117–127 [14] ASTM International. Standard test method for determination of ignition delay and derived cetane number DCN of diesel fuel oils by combustion in a constant volume chamber. ASTM D6890; 2008. [15] Yang SY, Naser N, Chung SH, Cha J. Effect of temperature, pressure and equivalence ratio on ignition delay in ignition quality tester (IQT): Diesel, nheptane, and iso-octane fuels under low temperature conditions. SAE Int J Fuels Lubr 2015;8(3):537–48. [16] Bogin G, Dean AM, Ratcliff MA, Luecke J, Zigler BT. Expanding the experimental capabilities of the ignition quality tester for autoigniting fuels. SAE Int J Fuels Lubr 2010;3(1):353–67. [17] Kuti OA, Yang SY, Hourani N, Naser N, Roberts WL, Chung SH, Sarathy SM. A fundamental investigation into the relationship between lubricant composition and fuel ignition quality. Fuel 2015;160:605–13. [18] Rosseel E, Sierens R. The physical and the chemical part of the ignition delay in diesel engines. SAE technical paper 961123; 1996. [19] Bogin Jr GE, Osecky E, Chen JY, Ratcliff MA, Luecke J, Zigler BT, Dean AM. Experiments and computational fluid dynamics modeling analysis of large nalkane ignition kinetics in the ignition quality tester. Energy Fuels 2014;28 (7):4781–94.
127
[20] Yang H, Shuai S, Wang Z, Wang J, Xu H. Performance of straight-run naphtha single-and two-stage combustion modes from low to high load. Int J Engine Res 2013;14(5):469–78. [21] Mittal V, Heywood JB, Green WH. The underlying physics and chemistry behind fuel sensitivity. SAE Int J Fuels Lubr 2010;3(1):256–65. [22] Zheng Z, Badawy T, Henein N, Sattler E. Investigation of physical and chemical delay periods of different fuels in the ignition quality tester. J Eng Gas Turb Power 2013;135:061501. [23] Cleveland WS, Devlin SJ. Locally weighted regression: An approach to regression analysis by local fitting. J Am Stat Assoc 1988;83(403):596–610. [24] Cannella W, Foster M, Gunter G, Leppard WR. FACE gasolines and blends with ethanol: Detailed characterization of physical and chemical properties. CRC report no. AVFL-24; 2014. p. 1–60. [25] Hildingsson L, Kalghatgi G, Tait N, Johansson B, Harrison A. Fuel octane effects in the partially premixed combustion regime in compression ignition engines. SAE paper, 2009-01-2648; 2009. [26] Ryan TW, Matheaus AC. Fuel requirements for HCCI engine operation. SAE paper, 2003-01-1813; 2003.