SAE 2015-01-0757
A Simple Method to Predict Knock Using Toluene, N-Heptane and Iso-Octane Blends (TPRF) as Gasoline Surrogates Gautam Kalghatgi, Hassan Babiker and Jihad Badra Saudi Aramco
ABSTRACT
1. INTRODUCTION
The autoignition resistance of a practical gasoline is best characterized by the Octane Index, OI, defined as RON-KS, where RON and MON are respectively, Research and Motor Octane Numbers, S is the sensitivity (RON-MON) and K is a constant depending on the pressure and temperature history of the fuel/air mixture in an engine. Experiments in knocking SI engines, HCCI engines and in premixed compression ignition (PCI) engines have shown that if two fuels of different composition have the same OI and experience the same pressure/temperature history, they will have the same autoignition phasing. A practical gasoline is a complex mixture of hydrocarbons and a simple surrogate is needed to describe its autoignition chemistry. A mixture of toluene and PRF (iso-octane + n-heptane), TPRF, can have the same RON and S as a target gasoline and so will have the same OI at any given K value and will be a very good surrogate for the gasoline.
The efficiency of SI engines is fundamentally limited by knock, an abnormal combustion phenomenon caused by the autoignition of the end-gas ahead of the flame front [1, 2, 3]. Knock depends on the pressure and temperature history of the end-gas as well as on the anti-knock quality of the fuel. The autoignition chemistry cannot be adequately quantified for practical fuels because of their complexity. Hence, fuel anti-knock quality has to be described by empirical measures and it is traditionally specified by Research and Motor Octane Numbers, RON and MON, of the fuel used. These tests are run in a single-cylinder CFR (Cooperative Fuels Research) engine in accordance with the procedures set in [4] – ASTM D2699 for RON and ASTM D2700 for MON. The RON test is run at an engine speed of 600 rpm and an intake temperature of 52° C while the MON test is run at 900 rpm and with a higher intake temperature of 149°C. The octane scale is based on two paraffins, n-heptane and isooctane. Blends of these two primary components are referred to as primary reference fuels (PRF) and define the intermediate points in the RON or MON scale. However, the autoignition chemistry of non-paraffinic components in gasoline differs from that of PRF and RON or MON describes the anti-knock behavior of the gasoline only at the RON or MON test condition. The true anti-knock quality of a gasoline is best described by an Octane Index, OI [3, 5, 6, 7] which is defined as
In this paper, a method to define the composition of a TPRF to match both RON and MON of a target gasoline is presented. The appropriate TPRF as a surrogate for a particular gasoline, which has been extensively tested in a knocking SI engine, is identified using this method. A chemical kinetic model is used to calculate ignition delays at different pressures and temperatures for this surrogate TPRF. From these data, a simple Arrhenius type equation with a pressure correction to predict ignition delays is identified. This equation is used to find the ignition delay as a function of crank angle and calculate the Livengood-Wu integral, I, for a number of individual knocking cycles covering a wide range of operating conditions using the gasoline in a single cylinder engine. Knock is predicted to occur at the crank angle when the integral, I, reaches unity. The crank angle at which knock is predicted to occur using the simple equation for ignition delay for the surrogate TPRF agrees very well with the experimentally observed value for the gasoline for all the cases considered. Finally, using the chemical kinetic model for TPRF, simple equations which can be used to estimate ignition delay are presented for a range of RON and sensitivity. Such equations can be used to predict when knock occurs during the cycle for these gasolines if the pressure and temperature development with crank angle is known.
OI = (1-K)RON + KMON = RON -KS
(1)
where K is an empirical constant which depends on the pressure and temperature history of the unburned mixture in the cylinder and S is the sensitivity (RON-MON). OI is the octane number of the PRF that matches the knock behavior of the gasoline that is of interest, at the conditions of the particular engine test. The temperature of the unburned gas increases as pressure increases in the engine cylinder and pressure and temperature in the end-gas are determined by engine design and operating conditions. Practical fuels have S > 0, typically around 10 and S can be taken as a measure of how different the fuel autoignition chemistry is compared to PRF, which by definition, have zero sensitivity. 1
The value of K has to be determined empirically by measuring a quantity dependent on fuel autoignition quality, such as the Knock Limited Spark Advance, KLSA, at a fixed operating condition for fuels of different RON and MON [3, 5, 6]. K is a measure of how different the pressure and temperature regime is compared to the RON test condition. If the temperature of the unburned gas for a given pressure is above that in the RON test condition K > 0; otherwise K < 0.
Shock tube experiments show that in the region where ignition delay is low enough to contribute significantly to the Livengood-Wu integral, n is much smaller for non-PRF fuels compared to PRF [3]. In other words, τi decreases less (knock resistance increases) for non-PRF fuels compared to PRF fuels as pressure increases for a fixed temperature. Multiple studies in knocking SI [5-8], HCCI [14,15] and PCI [16] engines show that if two fuels of different composition are exposed to the same pressure/temperature history and have the same OI, they will have the same autoignition phasing. The OI of practical fuels (S > 0) will change as the value of K changes, unlike for PRF. Hence PRF are not appropriate as surrogates for practical gasolines. A blend of toluene and nheptane, toluene reference fuel (TRF) which has the same RON as a practical gasoline would have a comparable, though not the same, sensitivity and would be a better surrogate than a PRF. If the scale in the anti-knock rating system was based on toluene and n-heptane, rather than PRF, a single “Toluene Number”, TN, the volume percent of toluene in the toluene/nheptane mixture, can describe anti-knock behavior of real gasolines reasonably well at different conditions, unlike RON or MON [17]. More importantly, such a system would also enable fuels such as ethanol mixtures, with higher anti-knock quality than iso-octane, to be rated quantitatively within the scale.
The value of K and hence OI for a given fuel depends on the design and operating condition of the engine. Measures aimed at improving the efficiency of a SI engine, such as increasing the compression ratio and turbocharging (aligned with downsizing) increase the maximum pressure and the temperature of the charge making autoignition and knock more likely. Fuel anti-knock quality becomes more important. Moreover, with increase in efficiency, in the unburned gas, for a given pressure, the temperature decreases or alternatively, for a given temperature, the pressure increases. This happens with turbocharging in down-sized engines and if the compression ratio is increased, because the amount of hot residual charge at the end of the exhaust stroke is reduced. In [5], Tcomp15, the unburned mixture temperature when the pressure is 15 bar (1.5 MPa) was introduced and helps to place the operating condition of interest in relation to the RON test condition. If Tcomp15 is less than that for the RON test, the condition is said to be “beyond RON” and K in Eq. 1 is negative so that for a given RON, a lower MON fuel has higher OI, i.e. higher anti-knock quality. Modern engines and downsized turbocharged SI engines of the next generation have negative values of K at the conditions where their performance is limited by knock [5-13] so that for a given RON, a fuel with greater sensitivity has more resistance to autoignition. Studies in HCCI engines have helped greatly in understanding autoignition and the dependence of K on Tcomp15 in pre-mixed mixtures [3].
However a proper surrogate should match both RON and MON of the gasoline. The simplest such surrogate fuel would require three components – iso-octane, n-heptane and toluene (TPRF fuels) [18]. Dryer and co-workers [19, 20] have also suggested that such fuels can be appropriate surrogates for gasoline. In this paper, a method to define the composition of a TPRF to match both RON and MON of a target gasoline is first presented. We then focus on experimental results on knock in a single cylinder engine running on a Saudi market gasoline with RON of 95.1 and MON of 86. The appropriate TPRF as a surrogate for this gasoline is first identified. A chemical kinetic model is used to calculate ignition delays at different pressures and temperatures for this surrogate TPRF. From these data, a simple Arrhenius type equation with a pressure correction to predict ignition delays (< 15 ms) is identified. This simple equation is used to find the ignition delay as a function of crank angle and calculate the Livengood-Wu integral, I, for a number of individual knocking cycles covering a wide range of operating conditions for which pressure was measured and the temperature was estimated. Knock is predicted to occur at the crank angle when the Livengood-Wu integral (Eq.3) reaches unity. The crank angle at which knock is predicted to occur using the simple equation for ignition delay for the surrogate TPRF agrees very well with the experimentally observed value for gasoline for all the cases considered. Finally, using the chemical kinetic model for TPRF, simple equations which can be used to estimate ignition delay are presented for a range of RON and sensitivity. Such equations
Thus, as the unburned mixture pressure increases with the temperature held constant (Tcomp15 decreases), the value of K decreases, and non-PRF fuels (S > 0) become relatively more resistant to autoignition compared to PRF. This observation is consistent with shock tube measurements of ignition delay for different types of fuels. The ignition delay, 𝜏𝑖 , can be expressed as a function of temperature, T, and pressure, P, in general as 𝜏𝑖 𝜏𝑖0
𝑃 −𝑛
= 𝑓(𝑇) ( ) 𝑃0
(2)
Here, τi0 is the ignition delay measured at a pressure of P0 at a given temperature, T, and n is a constant. In IC engines at each point during the cycle the mixture will have a different value of τi. It is then hypothesized that auto-ignition occurs when the integral (Livengood-Wu integral) of the reciprocal of τi with time, t, reaches unity (Eq. 3). 𝑡𝑒
𝐼 = ∫0
𝑑𝑡 𝜏i(𝑃,𝑇)
= 1
(3)
2
can be used to predict when knock occurs during the cycle for gasolines.
Also shown in Table 1 are TN from Eq.s 8 and 11 in [17]. TN is the volume percent of toluene in a toluene/n-heptane mixture which matches the fuel under consideration in the RON test (has the same RON).
2. DEFINING THE COMPOSITION OF A TPRF TO MATCH BOTH RON AND MON
In addition to these data, the data for TPRF with RON > 100 are available and are listed in Appendix A for completeness. For RON > 100, the position of the head in the RON test expressed as the DCR (Digital Counter Reading) can be found from Table A4.1 in [4]. Then using Eq. 4 from [17], the equivalent TN can be found and this is also listed in Appendix A.
Previous work in this regard was published by Morgan et al. [18] and Knop et al. [21]. However in the current paper the model is based on measurements of RON and MON over a more extensive range. The model equations are fundamentally correct in that the sensitivity goes to zero when the toluene concentration goes to zero. Model predictions agree extremely well with other data not used in developing the model
In general, even when blending rules for RON on a volumetric basis are non-linear, blending rules on a molar basis are linear [17, 21, 22]. Figure 1 shows measured RON plotted against TMF (data from Table 1) at different PRF levels.
The RON and MON values were measured for mixtures of PRF of different octane numbers with different concentrations of toluene. Blends with RON or MON greater than 100 were not considered for developing the model. The extension of the RON scale beyond 100 provided by ASTM is an extrapolation based on using octane enhancers within the scale e.g. by adding a known amount of lead to a lower octane PRF, measuring the increase in RON and assuming that the same increase in RON will be obtained if the same amount of lead is added to isooctane. However we cannot be certain that such an extrapolation is valid quantitatively for other fuels. For fuels with RON > 100, the actual head position of the CFR engine during the RON test is a physical measurement but the RON number associated with this is based on an extrapolation. A fuel with RON greater than 100 cannot be actually compared with a PRF since such a PRF cannot exist. Hence the current method to assign RON values to fuels that are more resistant to autoignition compared to isooctane, is not quantitatively reliable. If RON is greater than 100 we can say that the fuel is more resistant to autoignition compared to isooctane in the RON test but not quite by how much. In contrast, a TRF (toluene/n-heptane) can be found to match such a fuel in the RON test and a Toluene Number (TN) assigned to it [17].
120
RON
100
TMF =
20 0.1
100
=
𝑉𝑖 𝑉𝑖+𝑉𝑛
0.7
0.9
With a fixed PRF, as the toluene concentration is increased, RON increases linearly with TMF so that RON can be related to the toluene mole fraction, TMF, by 𝑅𝑂𝑁 = 𝑎𝑇𝑀𝐹 + 𝑏
(7)
Both the constants, a and b are listed in Table 2 and are different for each PRF level. Table 2 also lists the Rsq value for each linear fit shown in Fig.1. Also listed in Table 2 is the result for PRF zero, i.e. the blends of n-heptane and toluene from Eq.11 in [17] which was based on a large number of measurements. This equation, rather than Eq.9 in [17], has the right format in that with TMF = 0, RON is zero, the value for n-heptane. The constants a and b in Eq.7 are linear functions of PRF as shown in Fig.2 and represented in Eq.s 8 and 9. Once again the linear fit is excellent with Rsq values of 0.9968 and 0.9978 for a and b respectively. 𝑎 = 118.09 − 1.0355 𝑃𝑅𝐹
(8)
𝑏 = 1.0023𝑃𝑅𝐹 − 1.7774
(9)
From Eq.s 7, 8 and 9 we can write
(5)
𝑅𝑂𝑁 = (1.0023 − 1.0355𝑇𝑀𝐹)𝑃𝑅𝐹 + 118.09𝑇𝑀𝐹 − 1.777 (10)
Of course, 𝑉𝑖 + 𝑉𝑛 + 𝑉𝑡 = 100
0.5
Fig. 1. RON vs TMF at different PRF levels
PRF in Table 1 is the octane number of the PRF used so that 𝑃𝑅𝐹
0.3
TMF, Toluene Mole Fraction
(4)
𝐕𝐭+𝟎.𝟔𝟒𝟑𝟕𝐕𝐢+𝟎.𝟕𝟐𝟓𝟏𝟓𝐕𝐧
PRRF 80 PRF 20 PRF 40 PRF 60
60 40
The upper half of Table 1 (non-shaded) lists the data used to develop the model. Also listed in the shaded bottom half of Table 1 are data for TPRF with RON < 100, mostly from [4, 18, 19] which were not used in developing the model but were compared to the model – see below. In Table 1, Vi, Vn and Vt are respectively, the volume percentages of iso-octane, nheptane and toluene in the blend. TMF is the toluene mole fraction in the blend. In calculating TMF, the density (g/cc)/molecular weight used for toluene, iso-octane and nheptane are respectively 0.867/92.1, 0.692/114.2 and 0.684/100.2. With these assumptions, TMF is given by 𝐕𝐭
80
Figure 3 shows the sensitivity, S from Table 1 plotted against TMF. For PRF zero, Eq.s 11 and 12 from [17] give S = 13.9TMF. It can be seen that, not surprisingly, S seems to be
(6) 3
independent of PRF. The linear model forced to go through the origin is given by Eq. 11 and the Rsq value of the fit with
this model is 0.957. Equation 11 has the right format because Table 1
Experimental data for TPRF with RON < 100 The data in the non-shaded area used for developing the model to determine the composition of TPRF to match target RON and MON
Source Current Current Current Current Current [4] [4] Current Current Current Current Current Current Current Current Current Current Current Current Current Current Current [4] [4] [4] [4] Current Current [18] [18] [18] [18] [18] [21] [21] [21] [21] [21] [21] [21] [21] [21] [21]
Vi, vol % iso-oct
Vn, vol % n-hep
PRF
18 16 14 12 10 5 10 36 32 24 20 28 54 48 36 30 42 64 56 48 40 72 0 0 0 0 10 20 17 0 33 67 17 0 72 44 52 0 65 35 69 74 43
72 64 56 48 40 21 16 54 48 36 30 42 36 32 24 20 28 16 14 12 10 18 42 34 30 26 30 20 64 50 33 17 17 27 10 17 15 21 10 15 17 16 14
20 20 20 20 20 19 38 40 40 40 40 40 60 60 60 60 60 80 80 80 80 80 0 0 0 0 25 50 21.0 0.0 50.0 79.8 50.0 0.0 87.9 72.5 77.8 0.0 86.7 70.0 80.7 82.1 75.8
Vt, vol % tolu
10 20 30 40 50 74 74 10 20 40 50 30 10 20 40 50 30 20 30 40 50 10 58 66 70 74 60 60 19 50 33 16 67 73 18 40 34 79 25 50 14 10 44
TMF 0.136 0.261 0.377 0.485 0.585 0.800 0.804 0.138 0.265 0.490 0.591 0.382 0.141 0.270 0.496 0.597 0.388 0.275 0.394 0.503 0.602 0.144 0.656 0.728 0.763 0.797 0.680 0.687 0.251 0.580 0.423 0.229 0.744 0.792 0.250 0.500 0.436 0.840 0.337 0.599 0.201 0.140 0.537
MON
48.0 58.0 68.0 85.2 88.7
68.0 76.2 61.0 64.4 70.0 79.6 82.9 74.0 85.6 86.9 88.7 90.9 82.0 66.9 74.8 78.2 81.5 75.2 83.7 37.0 57.7 70.9 84.0 87.4 80.7 90.3 85.8 86.7 86.2 90.5 87.3 84.2 84.6 88.3
RON 32.0 42.0 53.2 63.7 75.5 96.9 99.8 48.0 58.0 75.1 83.8 66.1 66.0 73.6 86.2 92.1 79.0 89.1 92.8 96.7 99.8 84.5 75.6 85.2 89.3 93.4 85.3 95.0 39.0 65.9 76.2 87.0 98.0 92.3 93.7 93.0 93.0 97.7 95.2 96.3 86.6 85.7 96.3
S
5.2 5.7 7.5 11.7 11.1
7.1 7.6 5.1 1.6 3.6 6.6 9.2 5 3.5 5.9 8 8.9 2.5 8.7 10.4 11.1 11.9 10.1 11.3 2 8.2 5.3 3 10.6 11.6 3.4 7.2 6.3 11.5 4.7 9 2.4 1.1 8
*Data for TPRF with RON > 100 are given in Appendix A. Shaded numbers not used to develop model.
4
TN from Eq.s 11, 8 [17 ] 21.7 29.2 38.1 46.9 57.5 78.7 81.8 33.9 42.1 57.2 65.4 49.0 49.0 55.8 67.8 73.7 60.8 70.7 74.4 78.5 81.8 66.1 57.6 66.8 70.9 75.1 66.9 76.7 26.9 48.9 58.2 68.6 79.9 73.9 75.4 74.7 74.7 79.6 76.9 78.1 68.2 67.3 78.1
Table 2 Constants a and b in Eq. 7 from Fig. 1
PRF 0* 20 40 60 80
a 115.9 98.978 78.181 56.892 33.395
b 0 16.945 36.996 57.853 79.773
Figures 4 and 5 compare the predicted values using Eq.s 10 and 11, with the measured values for RON and MON respectively. There is excellent agreement between predicted and measured values (Rsq of 0.9998 for RON and 0.9976 for MON) even when cases which were not used to develop the model (the shaded numbers in the lower half of Table 1) are included.
Rsq 0.989 0.9976 0.9993 0.9973 0.995
*From Eq.11 in [17]
a
100 80
a (RON)
60
b (RON)
40 b = 1.0023PRF - 1.7774 R² = 0.9978
20 0 0
50
MON Model
Sensitivity
60 80 RON Measured
100
100
PRF 20 PRF 40 PRF 60 PRF 80 PRF 0, from [17] Model
0.5 TMF (Toluene Mass Fraction)
80 60
Current, used in model Current, not used in model Morgan et al [18] Knop et al [19] [4], not used in model
40 20 20
40
60 80 MON Measured
100
Fig.5 Predicted MON from Eq.s 10 and 11 vs Measured MON
1
Figure 6 shows the PRF needed in the TPRF blend vs the target sensitivity for different target RONs. Clearly with TPRF, maximum sensitivity is obtained when PRF is zero i.e. with toluene/n-heptane blends. The lower the target RON, the lower is the maximum sensitivity attainable. As we lower RON, the blend heads towards n-heptane.
Fig. 3. Sensitivity vs Toluene molar fraction for different PRF in TPRF blends when the toluene concentration is zero, the sensitivity should be zero regardless of the PRF value. 𝑆 = 14.07𝑇𝑀𝐹
40
Fig. 4 Predicted RON from Eq.10 vs Measured RON
Fig.2. The constants a and b (Eq. 7) plotted against PRF. Data from Fig.1, Table 2
0
Current, used in model Current, not used in model Morgan et al [18] Knop et al [19] [4], not used in model
20
100
PRF
16 14 12 10 8 6 4 2 0
100 90 80 70 60 50 40 30 20
RON Model
a = -1.0355PRF + 118.09 R² = 0.9968
120
90 80 70 60 50 40 30 20 10 0
b
140
3. A SIMPLE MODEL FOR IGNITION DELAY FOR A GASOLINE USING A TPRF SURROGATE
(11)
If we have a target RON and S, we first find the value for TMF using Eq. 11 and then, using Eq. 10, find the appropriate PRF for the blend. Then by solving Eq.s 4, 5 and 6, the individual volume percentages in the blend of iso-octane (Vi), n-heptane (Vn) and toluene (Vt) can be found.
A Saudi Arabian market gasoline with 95.1 RON and 86 MON was tested in a single cylinder port fuel injected (PFI) research engine at different operating conditions in [17]. The fuel contains 12.3% MTBE and further details about the fuel 5
can be found in [17] – also Appendix B. From the procedure described above, the TPRF that matches this gasoline for both RON and MON is made up of 55.3% vol toluene, 27.6% vol iso-octane and 17.1% vol n-heptane. 85 RON 88 RON 90 RON 92 RON 95 RON 98 RON 100 RON
100 80
PRF
coefficient) region. It has been suggested that the ignition delay should be modeled by three separate equations like Eq.12 in different temperature ranges [26]. However ignition delays at low temperatures and pressures are too large to contribute significantly to the Livengood-Wu integral and, as we show below, even though we completely ignore this regime, we can predict knock occurrence extremely well using a simple equation for ignition delay. Of course the kinetic model which yields the data in Fig.7 fully and necessarily includes the chemistry in this low temperature regime.
60
15 bar 20 bar 25 bar 30 bar 35 bar 40 bar 45 bar 50 bar 55 bar
20.0 τi, ms from Andreas et al [23]
40 20 0 6
7
8 9 10 11 12 Sensitivity = 14.07(TMF)
13
Fig.6 PRF needed in the TPRF blend vs sensitivity for different target RON values.
0.2
We first calculate the ignition delay, τi, for this TPRF at different pressures (P) and temperatures (T) at stoichiometric conditions using the chemical kinetic model described in [23] using the CHEMKIN software suite [24]. The results are shown in Fig.7. The data are also listed in Appendix B. We only consider ignition delays lower than 15 ms because we want to focus only on the regime that contributes significantly to the Livengood-Wu integral, Eq.3. For instance at an engine speed of 1500 RPM, 15 ms is equivalent to 135 crank angle degrees (CAD) so that a time step of one CAD contributes less than 1% of the target of unity, required to signal the occurrence of autoignition, to the Livengood-Wu integral. Of course, we can use another kinetic model to generate data such as considered in Fig. 7. Indeed, Appendix C lists data generated using a model described in [25]. It can be seen that it yields ignition delays that are larger than given by the model in [23]. We choose to go with the model from [23] – see discussion in Section 6. We then fit a simple Arrhenius type equation with a pressure correction of the form shown in Eq.12 to these data. 𝜏𝑖 = 𝐴𝑒𝑥𝑝(𝐵𝑇)𝑃−𝑛
2.0
0.8
1.0 1000/T, T in K
1.2
Fig.7. Ignition delays using the kinetic model from [23] at different pressures and temperatures for the TPRF surrogate (55%v toluene, 28% v iso-octane, 17% v, n-heptane) for a gasoline of 95.1 RON and 86 MON
τi ,from Eq. 12, ms
12.0 10.0
15 bar 20 bar 25 bar 30 bar 35 bar 40 bar 45 bar 50 bar 55 bar
8.0 6.0 4.0 2.0 0.0 0.0
5.0
10.0
τi , ms from kinetic model [23]
(12)
Fig.8. Ignition delays predicted by Eq.12 with A = 0.021682 ms, B = 7485 K and n = 1.125 compared with values in Fig.7. The solid line is the equivalence line.
This is done by fitting a linear model by multiple regression, with 1/T, and 𝑙𝑛(𝑃) as independent variables, and 𝑙𝑛(𝜏i ) as the dependent variable. For the data shown in Fig.7, such an exercise yields A = 0.021682 ms, B = 7485 K and with P expressed in bar, n = 1.125.
4. PREDICTION OF KNOCK PHASING USING EQ.12 AND COMPARISON WITH EXPERIMENTAL DATA FOR GASOLINE
Figure 8 compares the data in Fig.7 with the values predicted by Eq.12 and it can be seen that the simple model predicts the ignition delay reasonably well (Rsq of 0.992) over the pressure and temperature range considered. Clearly, a simple model such as Eq.12 will be wrong at low pressures and temperatures e.g., in the NTC (negative temperature
The gasoline (95.1 RON/86 MON) was tested in a single cylinder PFI (Port Fuel Injection) engine at seven different operating conditions listed in Table 3 – the first six of these 6
(Cond. 2 – Cond. 7) were also considered in [17]. At each of these operating conditions, the engine was run with different spark timings to establish the knock limited spark advance (KLSA) in [17] and the details about the engine and the fuel are in [17] but are reproduced in Appendix B. At each of these spark timings, 300 cycles of pressure data were collected at a resolution of 0.3 crank angle degrees. This resolution is coarser than used in some other knock studies but does not affect the main conclusions of the paper. We pick five or six knocking cycles at each operating point for further investigation. The criterion is that we should be able to clearly identify the point at which autoignition occurs as indicated by a sharp rise in pressure. There was another operating condition, Condition 1 in [17] but this is not considered in this work because the origin of autoignition could not be identified unambiguously. The spark timing used is also listed in Table 3 and Tcomp15 is calculated using the ideal gas law using the volume when the pressure is 15 bar and the estimated number of moles in the cylinder. The K values shown were calculated in [17] using KLSA for different fuels.
6.0
Pressure, MPa
5.0
-20 -15 -10
Intake Temp.˚ C
Intake Pr. KPa abs.
Spark Timing, CAD BTDC
Tcomp15 K
K value
2
1500
27
100
13
660
-0.12
3
1500
75
160
-4
645
4
1500
75
100
11
720
0.26
5
3000
27
160
12
615
-0.65
6
3000
27
100
26
660
0.1
7
3000
75
100
21
705
0.26
8
3000
75
160
10
652
5
10
15
20
25
30
We know the temperature, Tcomp15, of the bulk gas at 15 bar pressure during the compression stroke for each operating condition. This is calculated using the perfect gas equation (PV= mRT) after estimating the total number of moles, m, in the cylinder (air, fuel, residuals) because we know the volume in the cylinder when the measured pressure reaches 15 bar. However, the temperature is never homogeneous in an engine cylinder, primarily because of turbulent mixing of hot residuals with the cold charge, and autoignition leading to knock always starts at a hot spot [27, 3]. So we assume that when the pressure is 15 bar (P 0), T0 = (Tcomp15 +10) K for all cases. We then find the polytropic index, x, by plotting ln(P) vs ln(V) over 25 crank angle degrees before spark ignition i.e. towards the end of the compression stroke. Figure10 illustrates this for Condition 4 and x is found to be 1.31. We used this value for x for all the conditions considered. So the estimation of the temperature to be used in Eq.12 involves some uncertainties but the assumptions we have made are reasonable. As we show below, with these assumptions the observations are predicted very well. Figure 11 shows the ignition delay calculated for the six cycles shown in Fig.9 with these assumptions and Fig.12 shows the Livengood-Wu integral. The position of the occurrence of knock is given by the crank angle when this integral reaches unity and agrees with the position where there is a sharp change in pressure, indicating autoignition, in Fig.9, to within 1 crank angle degree.
Figure 9 shows pressure traces from six cycles at condition 4 as an example. At this condition all the cycles collected appear to be knocking – even the slower burning cycles such as Cycle 5. The cycles are different from each other, demonstrating cyclic variations. We estimate the temperature in the unburned gas in the following way. For adiabatic compression, the pressure, P, is related to the volume, V, by
Similar calculations were done for five cycles at each of the other operating conditions listed in Table 3. Table 4 summarises the predicted and observed crank angles when knock occurs in these cycles. In cycles which do not show knock, the Livengood-Wu integral does not reach unity over a very long crank angle window, if at all.
(13)
And if the temperature at a pressure P 0 is T0, the temperature, T, at any other given pressure, P, is given by 1
(𝑇/𝑇0 ) = (𝑃/𝑃0 )(1−𝑥)
1.0 -5 0
109 100 5
Fig.9. Pressure signals from six cycles at condition 4.
CAD BTDC – Crank angle degree before TDC
𝑃𝑉 𝑥 = constant
83 26 63
Crank Angle Degrees, ATDC
Conditions at which the gasoline was tested Speed , RPM
3.0 2.0
Table 3
Cond.
4.0
(14)
7
ln (P), P in bar
ln (P) = -1.3103ln (V) + 1.6326 R² = 0.9997
-10
Table 4.
14.6 14.4 14.2 14 13.8 13.6 13.4 13.2 13
-9.5 -9 ln (V), V in m3
Predicted and observed crank angles at knock origin Cond
No 2 2 2 2 2 3 3 3 3 3 4 4 4 4 4 4 5 5 5 5 5 6 6 6 6 6 7 7 7 7 7 8 8 8 8 8
-8.5
Ignition delay, ms
Fig. 10. ln (P) vs ln(V) between 46 CAD BTDC and spark ignition (11 CAD BTDC) for Cond. 4. The polytropic index, x, used in the temperature calculations is the negative of the slope of this line i.e. 1.31. 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 -20 -15 -10 -5 0
83 109 26 100 63 5
5
10 15 20 25 30
Crank Angle Degrees, ATDC
Observed Knock Occurrence Crank Angle, degrees ATDC
Fig.11. Ignition delay calculated from Eq.12 using measured pressure (Fig.9) and estimated temperature in the hot spot. Condition 4.
Livengood-Wu Integral
1.0 0.8 0.6 83 109 26 100 63 5
0.4 0.2
0.0 -20 -15 -10 -5 0
5
Cycle
35 63 144 31 3 79 144 9 155 66 83 109 26 100 63 5 69 83 155 115 74 92 61 74 115 31 56 111 63 128 149 104 111 9 24 85
10 15 20 25 30
Observed
CAD, ATDC 15.4 16.6 17.6 18.4 20.1 32.6 34.2 34.4 34.9 34.8 15.8 16.2 17.2 17.7 19.7 25.3 24.6 24.6 26.7 30 27.6 10.5 10.4 12.7 12.3 13.5 17.4 17.7 18.9 19.1 15.1 34 33.6 35.1 35.7 34.8
CAD, ATDC 15.0 16.2 16.9 17.7 18.6 33.3 34.8 34.8 35.1 35.1 15.3 15.9 17.1 17.4 20.1 25.8 25.2 26.0 27.3 30.0 28.2 10.2 11.1 12.9 12.0 12.9 18.1 18.6 19.5 19.8 14.7 34.5 34.5 36.6 36.4 36.0
Difference CAD -0.4 -0.4 -0.7 -0.7 -1.5 0.7 0.6 0.4 0.2 0.3 -0.5 -0.3 -0.1 -0.3 0.4 0.5 0.6 1.4 0.6 0 0.6 -0.3 0.7 0.2 -0.3 -0.6 0.7 0.9 0.6 0.7 -0.4 0.5 0.9 1.5 0.7 1.2
40 35 30 Cond 2 Cond 3 Cond 4 Cond 5 Cond 6 Cond 7 Cond 8 Equivalence
25 20 15 10 10
Crank Angle Degrees, ATDC
Predicted
20 30 40 Predicted Knock Occurrence Crank Angle Degrees ATDC
Fig.13. Comparison of observed and predicted crank angles at the start of knock for all operating conditions considered.
Fig.12. Livengood-Wu integral from ignition delays in Fig.11. Condition 4. 8
Figure 13 shows the observed crank angle plotted against the predicted crank angle at the start of knock listed in Table 4. There is very good agreement between the two – to within 1 crank angle degree. Thus the approach we have outlined in this section is an effective method to predict knock in an engine if the pressure and temperature histories of the unburned gas are measured or predicted.
Also, our focus here is on practical gasolines so that Table 5 essentially covers sensitivities between 8 and 11. However, there is much interest in fuels e.g. ethanol mixtures, which have RON values quoted to be nominally greater than 100. As we have just discussed, we cannot find the right TPRF surrogate for such fuels. However, if such fuels are tested in the RON test, they will have a Digital Counter Reading (DCR) indicating the compression ratio at knock associated with them (Table A4.1 in [4]) and we can assign a Toluene Number (TN), the volume percent of toluene in a TRF (toluene/nheptane) using Eq.4 in [17].
5. SIMPLE EQUATIONS FOR IGNITION DELAY FOR OTHER GASOLINES In Table 5 we list, for different target RON (88-100) and sensitivity (8-12), the appropriate TPRF composition found from the procedure described in Section 2. For each of these TPRF surrogates, we calculated the ignition delays at different pressures and temperatures using the kinetic model from [23] and found the constants (A, B and n) in Eq. 12 as described in Section 3; these constants are also listed in Table 5. If we know or can predict the pressure and temperature evolution in the engine, using the data in Table 5, we can assess whether the cycle will knock for a gasoline with the same RON and MON as the surrogate using the Livengood-Wu integral as described in the previous section.
Of course, we can also assign a RON but this has no physical meaning since such a PRF cannot exist and it is based on an extrapolation of the RON scale that might not be valid. Table 6 in [17] has measured results for many such fuels, including pure ethanol and pure MTBE which have TN values of 92.6 and 96.2 respectively (i.e. they matched these TRFs in the RON test). We suggest that for such gasolines, with RON > 100, and tested in the ASTM D2699 (RON) test, we use TRFs with the appropriate TN as surrogates (rather than TPRFs). The sensitivity of such gasolines will be high and comparable to the TRF surrogate. We generated the ignition delays for TRFs of different TN (> 82) using the kinetic model from [23] and found the constants, A, B and n as described in Section 3. These are listed in Table 6 and could be used to estimate ignition delays for full boiling range gasolines with quoted nominal RONs of greater than 100 (TN > 82).
The data shown covers RON up to 100 because the model to predict the TPRF surrogate composition is developed using mixtures with RON < 100. As discussed in Section 2 and in [17], values for RON above 100 are not quantitatively reliable. Table 5
Composition and constants in Eq.12 for TPRF surrogates for gasolines of different RON and sensitivity Surrogate Composition
Constants in Eq. 12
RON
MON
S
(R+M)/2
TN* From eq.s 8 &11 in [17]
Isooct
n-hep
toluene
vol%
vol%
vol%
88 88 88 90 90 90 92 92 92 95 95 95.1 95 98 98 98 100 100 100 100
78 79 80 80 81 82 81 82 84 84 85 86 87 87 88 89 88 89 90 91
10 9 8 10 9 8 11 10 8 11 10 9.1 8 11 10 9 12 11 10 9
83 83.5 84 85 85.5 86 86.5 87 88 89.5 90 90.55 91 92.5 93 93.5 94 94.5 95 95.5
69.6 69.6 69.6 71.6 71.6 71.6 73.6 73.6 73.6 76.7 76.7 76.8 76.7 79.9 79.9 79.9 82.0 82.0 82.0 82.0
8.0 18.8 28.9 10.8 21.5 31.5 2.1 13.6 34.1 6.6 18.0 27.6 38.1 11.1 22.4 32.7 1.7 14.2 25.3 35.5
28.5 26.1 23.9 25.9 23.5 21.4 25.9 23.3 18.9 21.8 19.3 17.1 15.1 17.6 15.2 13.1 17.6 14.7 12.5 10.5
63.5 55.1 47.3 63.3 54.9 47.1 72.0 63.1 47.0 71.7 62.7 55.3 46.8 71.3 62.4 54.2 80.6 71.0 62.2 54.0
9
A
B
n
ms
K
(P in bar)
0.03146 0.03147 0.03428 0.02663 0.02735 0.03065 0.02516 0.02236 0.02704 0.01719 0.01689 0.02168 0.02198 0.01173 0.01216 0.01403 0.01050 0.00883 0.00958 0.01136
7023 7035 6990 7209 7201 7134 7298 7403 7292 7694 7719 7485 7554 8103 8085 7992 8243 8402 8352 8236
1.125 1.137 1.150 1.123 1.135 1.149 1.107 1.121 1.147 1.103 1.116 1.125 1.145 1.098 1.110 1.124 1.077 1.091 1.105 1.121
lines to go through the (0, PRF) point, we get a = 114.86 1.0366 PRF. However this makes little difference to the surrogate compositions listed in Table 5. For instance, the composition of the surrogate for the Saudi gasoline changes only slightly, to 55.2% Toluene (from 55.3%), 16.5% nheptane (from 17.1%) and 28.3% (from 27.6%) isooctane by volume.
Table 6 A, B, n values (Eq.12) for gasolines with nominal RON > 100 or TN > 82 A TRF surrogate, toluene / n-heptane mixture is used rather than a TPRF. (These models might not be appropriate for model fuels which contain high levels of PRF). Pressure in Eq. 12 is expressed in bar. TN
DCR*
RON**
A, ms
B, K
n
84 86 88 90 92 94 96 98 100
949 981 1012 1042 1071 1099 1124 1150 1176
100.9 101.9 103.1 104.1 105.4 106.9 108.6 110.5 112.4
0.008475 0.006450 0.004605 0.003603 0.002429 0.001624 0.001050 0.000498 0.000399
8487 8776 9105 9271 9635 10002 10390 11159 11199
1.066 1.057 1.041 1.013 0.989 0.963 0.932 0.910 0.857
We then tested if the knock behavior of a gasoline is indeed predicted by the appropriate TPRF. To do this, we first found the correct TPRF surrogate that matches a Saudi Arabian market gasoline of 95.1 RON and 86 MON. This gasoline contains around 12% MTBE and has been tested in knock experiments in a single cylinder engine in [17]. We then found the ignition delays for the TPRF using a kinetic model from [23] at different pressures and temperatures. However, we only considered the regime where the ignition delay is less than about 15 ms because higher values of ignition delay do not contribute significantly to the Livengood-Wu integral – Eq.3. A simple equation of the form of Eq.12 was then found to fit the ignition delays derived from the kinetic model. For RON = 95.1 and MON = 86, the values of A, B and n in Eq.12 for the TPRF, the surrogate for the gasoline, are respectively, 0.02168 ms, 7485 K and 1.125 (with P expressed in bar).
*DCR – Digital Counter Reading in the RON Test indicating the compression ratio at knock, using Eq.4 in [17]. If the gasoline is tested and assigned a RON, its DCR can be found from Table A4.1 in [4] **The assigned value of RON for the relevant DCR from Table A4.1 in [4]. However, a TRF physically exists, unlike a PRF, to match these DCR values.
Measured pressure from individual knocking cycles from seven widely different operating conditions with the engine running on gasoline were then considered. The temperature to be used in Eq.12 as a function of crank angle was estimated for each cycle from the measured pressure. For each of the operating conditions, the bulk temperature in the unburned gas when the pressure was 15 bar (Tcomp15) was first estimated. However, because of turbulent mixing of the residuals with the cooler fresh charge, hot spots exist and autoignition always occurs at hot spots. To account for this we arbitrarily assumed that the temperature when the pressure was 15 bar was higher than Tcomp15 by 10 K. We then used the equations for adiabatic compression to calculate the temperature in the unburned gas at other pressures. The polytropic index in Eq.13 was obtained from the pressure/volume correlation near TDC before the spark fired and a common value of 1.31 was used for all cases. Thus there is some uncertainty in the estimation of temperature but the assumptions we make are reasonable.
However these data should not be used to describe model fuels with high levels of PRF. For instance, in Appendix A, the first three fuels are mixtures of high concentrations of iso-octane with toluene. These fuels match the toluene/n-heptane mixtures (TRF) with the relevant TN in the RON test. But we should expect those fuels to be more resistant to knock compared to the relevant TRF when K > 0 and more prone to knock when K < 0 (high pressure/low temperature conditions). Also listed in Table 6 is the relevant DCR in the RON test for each TRF using Eq.4 in [17] and the RON value for this DCR.
6. CONCLUDING DISCUSSION In this work we have developed a simple model which can be used to find the composition of a TPRF surrogate (a mixture of toluene, iso-octane and n-heptane) to match a given RON and sensitivity. It is based on experimental measurements of RON and MON of an extensive range of TPRF fuels. Such a surrogate should mimic the autoignition behavior of a gasoline of the same RON and sensitivity regardless of the pressure/temperature history in the engine cycle because the two fuels will have the same octane index as each other for all K values in Eq.1.
We then calculated the ignition delay using Eq.12 and the Livengood-Wu integral for the TPRF for all the knocking cycles and found that this approach predicted the position of knock occurrence with gasoline extremely well in all cases (Table 4 and Fig.13). Thus a TPRF which has the same RON and MON as the target gasoline was found to be a very good surrogate for the gasoline.
The model is very accurate and predicts the RON and MON of TPRF fuels not used in model development but available in the literature extremely well. The theoretically correct format for Eq.7 would require b = PRF. If we include this additional point (not tested) for TMF = 0 in Fig.1and force the best-fit
With this approach we are ignoring the low temperature and low pressure chemistry – indeed, a single equation of the form of Eq.12 would be plainly incapable of predicting ignition delays in these regimes. However, for engine applications, with the engine speed above 1000 RPM, if we can have a good 10
model to predict ignition delays which are lower than about 15 ms, it appears that knock occurrence can be predicted very well if we know the pressure and temperature in the unburned gas through measurement or modelling. Of course, the accuracy of the kinetic model itself, used to generate the ignition delay (e.g. Fig.7 and Appendix C) data in the first place is fully dependent on getting the low temperature / low pressure chemistry right.
combustion phasing, mixtures of PRF and toluene/ diisobutylene /and ethanol (20%vol) were considered. In [12] which considered knock in a single cylinder engine, again PRF mixtures with toluene/ diisobutylene /and ethanol (63.5% vol) were used. In [8], engine tests on knock considered 15 fuels including a blend with 8% MTBE. The engines used in these previous experiments included both PFI and DISI (direct injection spark ignition) engines. Hence we should expect that a TPRF fuel with the same RON and sensitivity (hence the same OI) will predict the autoignition phasing of the gasoline regardless of its composition.
We also present in Tables 5 and 6, the constants for Eq.12 which can be used to calculate the ignition delays of TPRF (and TRF) for a range of RON and sensitivities. These can be used to assess autoignition for gasolines that match these RON and sensitivities if the pressure in the unburned bulk gas is known as a function of crank angle. The temperature used in Eq.12 has to be estimated under some assumptions.
We have chosen to use the kinetic model from [23], which has been extensively validated, to generate the initial ignition delay data for TPRFs. However different kinetic models could be tried. In the present case, if we had used the kinetic model from [25], which shows much higher ignition delays (Appendix C), the predicted phasing for knock would be much more retarded than observed for all the cases. Indeed, in some cases, even though cycles were observed to knock, the prediction would have been that there would be no knock. One way to make the predictions using the model from [25] agree with the observations for the knock occurrence angle for cycles shown in Fig.9 is to assume that the temperature in the hot spot at 15 bar pressure is higher than the bulk temperature by 50 K rather than 10 K used in the current study. Such an assumption does not appear realistic.
The data in Tables 5 and 6 can be used to quickly assess the impact of fuels on engines. For instance, moving from the gasoline used in the current experiments to a fuel with 92 RON and 82 MON (U.S. Regular) will advance the predicted knock occurrence in Cycle 83 and Cycle 5 in Fig. 9 by about 0.7 and 2.0 crank angle degrees respectively using the constants in Table 5. On the other hand, if a fuel which has been assigned a RON of 105.4 in the RON test is of interest, the constants for TN 92 in Table 6 can be used. In that case the predicted knock occurrence for Cycle 83 will be retarded by 4 crank angle degrees compared to the gasoline considered in this paper whereas Cycle 5 will not knock at all (the Livengood-Wu integral does not reach unity).
There might be differences in low-temperature and low pressure chemistry between a gasoline, say with high levels of ethanol, and its TPRF surrogate. These differences might be highlighted if the two fuels are exposed to different initial conditions and pressure/temperature histories while holding the autoignition phasing during the cycle constant as in some HCCI studies. Of course, chemical kinetic models have to get such chemistry right in order to predict the ignition delays at higher pressure and temperature which contribute to the Livengood-Wu integral in engine studies. However, once this is achieved, it appears that the ignition delays below about 15 ms can be predicted well with much simpler models as in Eq. 12 and knock prediction in engines simplified.
Of course, the suitability of using the data in Tables 5 and 6 to predict knock can be properly assessed only through validation experiments such as the ones described in Section 4 using different gasolines of varying composition (e.g., those containing ethanol) and different engine configurations and injection systems (e.g., direct injection). There might be differences between the gasoline and its TPRF surrogate in terms of charge cooling and burn rates which affect knock in an engine. However such differences are likely to be less important compared to getting the autoignition behavior matched. The approach described works very well, at least for one gasoline, the one considered in this work which contains 12% MTBE. Such an approach has also been shown to work for the prediction of knock phasing in two different engines for a different fuel though only four knocking cycles were considered in that work [29].
ACKNOWLEDGMENTS The authors would like to acknowledge the contribution of their erstwhile colleague, Syed Sayeed Ahmed who organized the RON/MON measurements of the TPRF fuels. Our colleague, Yoann Viollet kindly made available the pressure curves for individual cycles used in Section 3.
The OI approach itself has been very extensively tested in knocking SI engines, HCCI engines and PCI engines [3]. Autoignition phasing is very well predicted by OI if the fuel/air mixtures are subjected to the same pressure and temperature history regardless of whether the sensitivity of gasolines arises from olefins, aromatics or oxygenates. For instance in [7] which considered knocking in vehicles, fuels including MTBE at up to 15% vol were considered. In [28] which took CA50 in HCCI engines as a measure of
CONTACT INFORMATION Gautam Kalghatgi, Principal Professional, Saudi Aramco, PO Box 62, Dhahran 31311, Saudi Arabia. Tel: +966 38768171 Email:
[email protected] 11
17. Kalghatgi, G., Head, R., Chang, J., Viollet, Y, Babiker, H and Amer, A., “An Alternative Method Based on Toluene/n-Heptane Surrogate Fuels for Rating the Antiknock Quality of Practical Gasolines”, SAE Paper 201401-2609.
REFERENCES 1.
Heywood, J.B. Internal combustion engine fundamentals. New York: McGraw Hill Book Co., 1988.
2.
Stone, R. Introduction to Internal Combustion Engines. 4th Ed. Basingstoke: Palgrave Macmillan, 2012.
3.
Kalghatgi, G.T., Fuel/Engine Interactions. Warrendale, PA.:SAE International, 2014
4.
Annual book of ASTM Standards Vol. 5.01-5.05. American Society of Testing Materials
5.
Kalghatgi, G.T., 2005, “Auto-ignition quality of practical fuels and implications for fuel requirements of future SI and HCCI engines”. SAE Paper 2005-01-0239.
6.
Kalghatgi, G.T., 2001, “Fuel Anti-knock Quality - Part I, Engine studies.” SAE Paper 2001-01- 3584
7.
Kalghatgi, G.T. Fuel anti-knock - Part II., 2001, “ Vehicle studies- how relevant is Motor Octane Number (MON) for modern engines?”, SAE Paper 2001-01-3585
8.
Kalghatgi, G.T., Nakata, K. and Mogi, K., 2005, “Octane appetite studies in direct injection spark ignition (DISI) engines.” SAE 2005-01-0244
9.
18. Morgan, N., Smallbone, A., Bhave, A., Kraft, M., Cracknell, R., Kalghatgi, G. 2010. “Mapping surrogate gasoline compositions into RON/MON space.” Combustion and Flame157: pp 1122-113
Ch.4.
19. Dryer, F., 2014, “Chemical Kinetic and Combustion Characteristics of Transportation Fuels”, Proc. Comb. Inst., Vol 35, in press. doi:10.1016/j.proci.2014.09.008 20. Chaos, M, Zhao, Z, Kazakov, A., Gokulkrishnan, P, Angioletti, M, Dryer, F., 2005. “A PRF+Toluene Surrogate Fuel Model for Simulating Gasoline Kinetics”, 5th U.S. Combustion Meeting, March 25-28, 2007. 21. Knop, V, Loos, M., Pera, C. and Jeuland, N., 2014. “ A linear-by-mole blending rule for octane numbers of nheptane/iso-octane/toluene mixtures” Fuel 115: pp 666673 22. Anderson, J.E., Kramer, U., Mueller, S.A. and Wallington, T.J., 2010. “Octane numbers of ethanol- and methanol-gasoline blends estimated from molar concentrations” Energy Fuels 24; pp 6576-6585
Amer, A., Babiker, H., Chang, J., Kalghatgi, G., Adomeit, P.,Brassat, A. and Guenther, M., 2012, “Fuel Effects on Knock in a Highly Boosted Direct Injection Spark Ignition Engine”, SAE Paper 2012-01-1634
23. Andrae, J., Brinck, T. and Kalghatgi, G.T. “HCCI experiments with toluene reference fuels modeled by a semi-detailed chemical kinetic model”, Combustion and Flame, vol 155, pp 696-712, 2008
10. Bell, A., 2010, “Modern SI Engine Control Parameter Responses and Altitude Effects with Fuels of Varying Octane Sensitivity”, SAE Paper 2010-01-1454, 2010.
24. Reaction Design, in: 15112 ed.; Reaction Design: San Diego, 2011
11. Fuel Antiknock Quality – Engine Response to RON vs MON, Scoping Tests – Final Report Coordinating Research Council (CRC), May 2011
25. Mehl M., Pitz W.J., Sarathy S.M., Yang Y., Dec, J.E. 2012 "Detailed Kinetic Modeling of Conventional Gasoline at Highly Boosted Conditions and the Associated Intermediate Temperature Heat Release", SAE Paper 2012-01-1109
12. Mittal, V. and Heywood, J.B., 2008, “The Relevance of Fuel RON and MON to Knock Onset in Modern SI Engines”, SAE Paper 2008-01-2414 13. Davies, T., Cracknell, R., Lovett, G., Cruff, L. and Fowler, J. 2011. “Fuel effects in a boosted DISI engine”, SAE Paper 2011-01-1985. Also JSAE 20119155
26. Yates, A.D., Swarts, A. and Viljoen, C.L. 2005. “Correlating Auto-Ignition Delays And Knock-Limited Spark-Advance Data For Different Types Of Fuel.” SAE Paper No. 2005-01-2083
14. Risberg, P., Kalghatgi, G.T. and Ångström H-E. 2003. “Autoignition quality of gasoline-like fuels in HCCI engines.” SAE Paper 2003-01-3215.
27. Bradley, D.2012. “Autoignitions and detonations in engines and ducts." Phil.Trans.R.Soc A. 370: pp 689-714 28. Kalghatgi, G.T., Risberg, P. and Ångström, H-E, “A method of defining ignition quality of fuels in HCCI engines.” SAE Paper 2003-01-181
15. Kalghatgi, G.T. and Head, R.A. 2004 “The available and required autoignition quality of gasoline-like fuels in HCCI engines at high temperatures” SAE Paper 2004-011969
29. Kalghatgi, G.T., Bradley, D., Andrae, J., Harrison, A.J., “The nature of “superknock” and its origins in SI engines”, I.Mech.E. Conference on Internal combustion engines: Performance, Fuel Economy and Emissions, in London, December 8-9, 2009, Chandos Publishing, 2009, ISBN 9781843346074
16. Kalghatgi, G.T., Hildingsson, L., Harrison, A.J., L. and Johansson, B., 2011. “Surrogate fuels for premixed combustion in compression ignition engines”, International Journal of Engine Research, vol. 12, 5: pp. 452-465 12
Appendix A Experimental data for TPRF with RON >100 Source
Vi
Vn
PRF
Vt
TMF
Current Current Current Current Current Current Current Current Current Current Current Current Current Current Current Current [4] [4] [18]
90 80 70 60 50 30 40 4 8 12 18 20 0 2 4 6 15 20 17
0 0 0 0 0 10 0 16 12 8 2 0 10 8 6 4 11 6 17
100 100 100 100 100 75 100 20 40 60 90 100 0 20 40 60 58 77 50
10 20 30 40 50 60 60 80 80 80 80 80 90 90 90 90 74 74 67
0.147 0.280 0.400 0.509 0.608 0.693 0.700 0.849 0.852 0.855 0.860 0.861 0.925 0.927 0.929 0.930 0.808 0.811 0.744
MON
RON
100.3 91.2 100.4 90.5 94.0 97.5 101.0 102.8 100.0 101.4 102.4 104.4 92.6 96.6 99.3
102.0 104.1 105.6 107.7 108.2 101.6 110.0 101.0 103.1 105.4 108.5 112.6 106.0 108.0 109.5 111.8 103.3 107.6 110.0
S
DCR from Table A4.1 [4]
TN from Eq. 4 [17 ]
983 1042 1074 1111 1118 969 1145 950 1014 1070 950 1179 1081 1115 1138 1167 1019 1110 1145
86.1 90.0 92.2 94.9 95.5 85.3 97.6 84.1 88.1 91.9 84.1 100.5 92.7 95.3 97.1 99.5 88.5 94.9 97.6
7.9 10.4 9.6 10.5 9.1 7.9 7.5 9.8 6 6.6 7.1 7.4 10.7 11 10.7
APPENDIX B Details about the fuel and engine used Engine Number of Valves Displacement, cm3 Bore, mm Stroke, mm Compression Ratio, Fuel Pump/ Range CR Fuel Injector Intake System Port fuel injection
Details of the Saudi Arabian gasoline used in the tests
4 499 84 90 10.5 External motor driven/ 4-15 Outwardly opening MPa Conditioned piezoelectric air with external boost
Naphthenes Paraffins i-Paraffins Aromatics Oxygenates (MTBE) olefins MON RON Density @ 15 C RVP ASTM D323 SULFUR D-5453 0-IBP 05% Recovery 10% Recovery 20% Recovery 30% Recovery 40% Recovery 50% Recovery 60% Recovery 70% Recvoery 80% Recovery 90% Recovery 95% Recovery End Point
13
2.8 15.3 29.5 37.4 12.3 2.6 86 95.1 0.7505 12.64 10.0 38.3 42.2 45.0 55.0 60.6 68.3 73.3 103.9 112.8 129.4 147.8 160.0 173
Vol% Vol% Vol% Vol% Vol% Vol% gm/cc psi ppm °C °C °C °C °C °C °C °C °C °C °C °C °C
Appendix C Ignition delays at stoichiometric conditions for different pressures and temperatures using two different chemical kinetic models for the TPRF representing a gasoline of 95.1 RON and 86 MON Temperature (K)
Pressure (bar)
Ignition Delay, ms, Andrea et al.,[23]
Ignition Delay (ms)-LLNL [25]
Temperature (K)
Pressure (bar)
Ignition Delay, ms, Andrea et al.,[23]
Ignition Delay (ms)-LLNL [25]
800
15
12.171
21.508
850
35
2.755
5.179
850
15
8.450
16.008
900
35
1.957
3.546
900
15
5.933
9.857
950
35
1.347
2.096
950
15
3.618
5.469
1000
35
0.821
1.172
1000
15
1.970
3.099
1050
35
0.473
0.676
1050
15
1.078
1.805
750
40
7.280
7.893
1100
15
0.621
1.000
800
40
3.721
5.517
800
20
8.458
14.317
850
40
2.325
4.335
850
20
5.742
10.910
900
40
1.640
3.014
900
20
4.094
6.966
950
40
1.142
1.808
950
20
2.618
3.919
1000
40
0.710
1.015
1000
20
1.476
2.199
1050
40
0.413
0.583
1050
20
0.818
1.288
750
45
6.483
7.062
1100
20
0.474
0.739
800
45
3.263
4.730
45
2.007
3.705
750
25
11.746
13.142
850
800
25
6.435
10.473
900
45
1.404
2.609
850
25
4.269
8.107
950
45
0.986
1.587
900
25
3.055
5.325
1000
45
0.623
0.896
950
25
2.019
3.046
1050
45
0.367
0.513
1000
25
1.173
1.701
750
50
5.851
6.428
1050
25
0.659
0.992
800
50
2.908
4.135
1100
25
0.383
0.580
850
50
1.762
3.220
50
1.223
2.292
750
30
9.723
10.650
900
800
30
5.179
8.137
950
50
0.863
1.412
850
30
3.362
6.359
1000
50
0.554
0.802
900
30
2.401
4.274
1050
50
0.330
0.458
950
30
1.624
2.486
750
55
5.337
5.930
1000
30
0.969
1.387
800
55
2.624
3.672
1050 750
30 35
0.551 8.316
0.805 9.026
850
55
1.569
2.837
900
55
1.080
2.037
950
55
0.765
1.270
1000
55
0.497
0.725
1050
55
0.299
0.414
800
35
4.330
6.595
14
