Topic: Internal Combustion and Gas Turbine Engines

Transcription

1 Paper # 7IC-264 Topic: Internal Combustion and Gas Turbine Engines 8 th U. S. National Combustion Meeting Organized by the Western States Section of the Combustion Institute and hosted by the University of Utah May 19-22, 213 Autoignition Characterization of Primary Reference Fuels and n- Heptane/n-Butanol mixtures in a Constant Volume Combustion Device and Homogeneous Charge Compression Ignition Engine Marc E. Baumgardner 1 Anthony J. Marchese 1 S. Mani Sarathy 2 1 Department of Mechanical Engineering, Colorado State University, Fort Collins, CO 2 Clean Combustion Research Center, King Abdullah University of Science and Technology, Thuwal, Kingdom of Saudi Arabia Premixed or partially premixed compression ignition modes, such as homogeneous charge compression ignition (HCCI), have been a particular focus among researchers because of their potential to deliver enhanced fuel efficiency and meet exhaust emissions mandates without the addition of costly after-treatment technologies as currently required with traditional spark ignition (SI) and direct injection compression ignition (DICI) engines. These advanced combustion strategies thus seek to combine the advantages of SI and DICI engines while avoiding their disadvantages. Previous studies have suggested that future fuels with desired properties for optimal performance in these advanced combustion modes might have properties between that of traditional gasoline and diesel fuels and will most likely consist of mixtures of petroleum derived products (aromatics, straight and branched alkanes), alcohols, synthetic alkanes, and fatty acid methyl esters. Existing ignition quality metrics such as Octane or Cetane Number are reasonably consistent for petroleum-derived fuels in SI and CI engines but can fail to adequately characterize autoignition in advanced combustion modes, particularly when higher concentrations of alcohols are included in the blend. Accordingly, in this study, the autoignition behavior of primary reference fuels (PRF) and blends of n-heptane/n-butanol were examined in a Waukesha Fuel Ignition Tester (FIT) and a Homogeneous Charge Compression Engine (HCCI). Fourteen different blends of iso-octane, n-heptane, and n-butanol were tested in the FIT 28 test runs with 25 ignition measurements for each test run, totaling 35 individual tests were done in all. These experimental results supported previous findings that fuel blends with similar octane numbers can exhibit very different ignition delay periods. The present experiments further showed that n-butanol blends behaved unlike PRF blends when comparing the autoignition behavior as a function of the percentage of low reactivity component. These same fuel blends were also tested in a John Deere 424T diesel engine, which was modified to operate in HCCI mode. The HCCI and FIT experimental results were compared against multi-zone models with detailed chemical kinetic mechanisms for PRF s and n-butanol. For both the FIT and HCCI engine data, a new n-butanol/n-heptane kinetic model is developed that exhibits good agreement with the experimental data. The results also suggest that that the FIT instrument is a valuable tool for analysis of high pressure, low temperature chemistry and autoignition for future fuels in advanced combustion engines. 1. Introduction Depletion of fossil fuels and climate change from anthropogenic greenhouse gas emissions arguably represent the first civilization-scale challenges ever faced by the human race (Hirsch, et al., 25) (McKibben, 26). While there is no single technological solution for these global challenges, an effective means of rapidly addressing these concerns is to increase the thermodynamic efficiency of energy conversion devices that consume fossil fuels. Because of its widespread use as a mobile energy source, the internal combustion engine will continue to be a principal source of greenhouse gas emissions and the principal consumer of liquid fossil fuels. As detailed in the 27 IPCC report on climate change (Pachauri, et al., 27), the global transportation sector accounts for 13.1% of the total anthropogenic contribution to greenhouse gasses (GHG). In the United States, the EPA estimates that 27% of the over 68 Tg of CO2 Eq emitted in 21 were from the transportation sector (EPA, 212), and 62% of these emissions were from (LDV) light duty vehicles (i.e. passenger cars and light duty trucks, SUV s, and minivans). Therefore, increasing the efficiency of internal combustion engines, presents an immediate and major opportunity for greenhouse gas reduction and extension of fossil fuel reserves. Spark Ignited (SI) engines, despite having increasingly lower emissions due to the advancements of 3-way catalysts are limited by the narrow range of air/fuel ratios required for 3-way catalyst operation and low compression ratios

2 required to prevent engine knock. The ultimate efficiency of an SI engine is thus restricted by this operationally low compression ratio and equivalence ratio range. Alternatively, compression ignition (CI) engines (i.e. diesel engines) operate at higher compression ratios and have lower pumping losses due to fuel injection near top dead center (TDC), which results in higher efficiency. However, the shortened fuel/air mixing time experienced in CI engines results in a more heterogeneous fuel/air mixture and higher local flame temperatures, which can result in high concentrations of oxides of nitrogen (NO x ) and particulate matter (PM). As a result, despite the benefits of high efficiency, CI engines have historically been the heaviest polluters of NO x and PM, both of which are highly problematic in terms of atmospheric pollution and associated public health issues. The urgent need to increase efficiency and reduce exhaust emissions from internal combustion engines has resulted in an increased interest in High Efficiency Clean Combustion (HECC) engines. Premixed or partially premixed compression ignition modes, such as homogeneous charge compression ignition (HCCI), reactivity controlled compression ignition (RCCI), and multi-zone stratified compression ignition (MSCI) have been a particular focus because of their potential to deliver enhanced fuel efficiency and meet exhaust emissions mandates without the addition of costly after-treatment technologies (Lu, et al., 211). These advanced combustion strategies thus seek to combine the advantages of SI and CI engines while avoiding their disadvantages. The effect of local equivalence ratio and temperature of these advanced combustion engine modes and their effect on NO x and PM formation are shown in Figure 1, along with a comparison to conventional CI engines (Neely, et al., 25). Despite the promise of these relatively new engine combustion modes, most have only been successfully demonstrated in steady state, lab-scale demonstrations with traditional liquid fuels (gasoline, diesel or primary reference fuels). Those that have been developed beyond the lab scale still suffer from restricted range and require advanced control strategies to allow switching between advanced (HCCI) and traditional (e.g. SI) operation (Yun, et al., 211). However, due to the many advantages of HCCI and other advanced combustion strategies, the U.S. government has long considered these research areas of paramount importance in achieving future emission standards and goals (DOE, April, 21). Many of the advanced internal combustion engine concepts under consideration, such as HCCI, RCCI, MSCI and other similar strategies, rely on a premixed or partially premixed autoignition event, which is typically governed by low temperature chemical kinetics. Many researchers have shown that combustion strategies such as HCCI or MSCI are more amenable to moderately reactive fuels (i.e. low to moderate Octane Number) (Bessonette, et al., 27) (Ciatti, 212) (Dec, 29) (Kaneko, et al., 22). However, research has also shown that existing autoignition metrics such as octane number (ON) or cetane number (CN) are not directly applicable toward characterizing autoignition under homogeneous premixed or partially premixed conditions (Bessonette, et al., 27) (Koopmans, et al., 24) (Perez, 212) (Liu, et al., 29) (Rapp, et al., 212). Accordingly, new fuel characterization methods and autoignition metrics must be developed that are applicable to future fuels in advanced combustion modes. Recently, there have been several approaches to characterizing fuel performance and rank in HCCI (and other HECC engine modes). These approaches have produced various levels of success but ultimately do not adequately address the characteristics of future fuels. One of the first major steps to identifying a common scale with which to examine potential HCCI fuels was a study by Ryan et al. (Ryan, 22). This Elevated Pressure Auto-Ignition Temperature (EPAIT) approach was based on data from an Ignition Quality Tester (IQT ). The IQT is a similar device to the Fuel Ignition Tester (FIT ) described herein and has its own ASTM standard for calculating DCN (D689, 212). The IQT directly injects fuel into a constant volume vessel that is initially at ±.7 bar and within a temperature range of 788 to 848K (Haas, et al., 211). In the EPAIT method, several fuels of interest are chosen and IQT experiments are conducted over a range of initial temperatures to obtain ignition delay period as a function of initial temperature. Next, an ignition delay period common to all fuels is chosen and the fuels are then ranked by their corresponding ignition temperatures at this common ignition delay time. Ryan et al. (22) showed a fair correlation with Start of Reaction (SOR) in an HCCI engine. In the EPAIT method, SOR is analogous (but not equivalent) to CA5. A major limitation of this method is that it is only 2 Figure 1. Effect of local equivalence ration on NO and PM formation in traditional and advanced combustion modes for internal combustion engines (Neely, et al., 25).

3 applicable for fuels that have similar ignition delay periods in an IQT. This limitation has been identified by others (Perez, 212) and based on the results presented herein this method would be inadequate to implement for a wide range of fuels. The Octane Index (OI) was developed by Kalghatgi (Kalghatgi, et al., 23) based on previous work studying knock in SI engines (Kalghatgi, 21) (Kalghatgi, 21). The OI is a measure of the propensity of a given fuel to knock based on the Sensitivity, S, of the fuel as defined by Equation 1: S=RON-MON (1) The Sensitivity parameter is said to be related to the chemical properties of a given fuel, which result in different temperature/pressure histories. More formally, S is considered to be a measure of the octane depreciation that occurs as the engine test speed is increased (Mehl, et al., 26). The antiknock properties of gasoline-type fuels (i.e. real fuels with aromatics, and alkenes, as well as alkanes) tend to lessen as engine RPM increases. It should be noted that primary reference fuels (PRF), upon which the RON and MON scales are based, are alkanes and so have fundamentally different ignition chemistry than aromatics, alcohols and other fuels. By definition, the sensitivity of an n-heptane/iso-octane PRF mixture is zero. In theory, the behavior of a fuel with respect to the RON and MON scales should give some indication of the alkane-likeness of its combustion kinetics. Additionally, since the RON test has a lower compression temperature for a given cylinder pressure (due mostly to the lower intake temperature of the RON test) some degree of delineation should be obtained by comparing these two measures. However, both the RON and MON tests are performed at speeds < 1 RPM, which are well below typical modern engine speeds under load. And, the RON and MON scales are inherently limited by being defined by two alkanes (i.e. i-octane and n-heptane). Kalghatgi tried to correct for this issue by introducing an empirical parameter, K, resulting in the following equation for Octane Index: OI=(1-K)RON+K(RON-MON) (2) It should be noted, however, that the parameter K is specific to a given engine and operating condition. The specificity of K severely limits the applicability of this index to real fuels across a wide array of engines, but the concept that the temperature and pressure history greatly impact the combustion timing is one that will be further addressed below. Though the OI represents a substantial development in trying to understand and rank real fuels under HCCI operating conditions, several studies have found that this method ultimately does not correlate well with CA5 when a broader range of fuels is considered (specifically, for example, when alcohols and other non-petroleum derived fuels are considered (Liu, et al., 29) (Perez, 212) (Rapp, et al., 212)). In 27, Shibata and Urushihara (Shibata & Urushihara, 27) built upon the Kalghatgi work by further differentiating the contributions of low temperature heat release (LTHR) and high temperature heat release (HTHR) to RON and MON as follows: OI=(1-K) LTHR+HTHR (3) In this regard, Shibata and Urushihara were (to the authors knowledge) the first investigators to attempt to rank fuel performance in an HCCI engine by focusing specifically on the LTHR component of the total heat release. Specifically, they examined 23 different model fuels made from 11 pure component hydrocarbons and tried to correlate the CA2 to an HCCI Index, where the HCCI Index was found by summing the individual component contributions to heat release as follows: HCCI Index abs =rron+a np +b ip +c O +d A +e OX +Y (4) where np, ip, O, A, and OX refer to fractions of n-alkanes, iso-alkanes, olefins, aromatics, and oxygenates, respectively, and r, a, b, c, e, and Y are empirical parameters (Shibata & Urushihara, 27). The major conclusions of this study were that paraffins were the main contributors to LTHR; aromatics inhibited LTHR; and olefins and naphthenes exhibited both effects. The HCCI Index method was effective for the fuels originally tested but has since proven less accurate for other fuels with similar ON (Perez, 212) (Rapp, et al., 212). Ultimately, the HCCI index suffers from similar disadvantages as OI since it is based on the RON and MON tests and is specific to a given engine and operating conditions. In recent years, the IQT has seen increased use as a fundamental combustion research tool because of its ability to rapidly measure the ignition delay of a variety of fuels at low to intermediate temperature (Bogin, et al., 211) (Bezaire, et al., 21) (Haas, et al., 211) (Perez, 212). Bogin et al. (Bogin, et al., 213) have recently shown that ignition delay 3

4 times for various heptane isomers tested in an IQT show similar behavioral trends as the same fuels tested in a homogenous rapid compression machine (RCM) by Silke et al (Silke, et al., 25). Bogin et al. modeled the IQT with KIVA-3V and reduced chemical kinetic mechanisms and concluded that the ignition behavior was dominated by chemical kinetic effects as compared to spray physics. Bogin also suggested that the dominance of chemical kinetics increased substantially for fuels with longer ignition delay periods (Bogin, et al., 211). Perez et al. further showed that fuels possibly best suited for HCCI such as those with a RON of approximately 7 exhibit IQT ignition delay periods of greater than ~8 ms, which is sufficiently long for the system to be modeled homogeneously (Perez, 212). Perez and coworkers tested a wide range of gasoline-like fuels to determine if HCCI autoignition behavior could be predicted or correlated to that measured in an IQT. They tested 21 different surrogate gasoline fuels in an IQT and an HCCI engine and found that the inverse of ignition delay period correlated relatively well with the CA5 data from the HCCI engine. Perez and coworkers also found that, while the inverse of ignition delay periods generally correlated with the RON and OI, there were substantial differences between fuels of the same RON that could not be explained by either the RON or OI scales. In this study, the autoignition behavior of primary reference fuels (PRF) and blends of n-heptane/n-butanol were examined in a Waukesha Fuel Ignition Tester (FIT) and compared against experiments in a Homogeneous Charge Compression Engine (HCCI). Fourteen different blends of iso-octane, n-heptane, and n-butanol were tested in the FIT. These experimental results presented herein support previous findings that fuel blends with similar ON can exhibit very different ignition delay periods. The results further suggest that n-butanol blends behave unlike PRF blends when comparing the autoignition behavior as a function of the percentage of low reactivity component. The HCCI and FIT experimental results were compared against multi-zone models with detailed chemical kinetic mechanisms for PRF s and n-butanol. For both the FIT and HCCI engine data, a new n-butanol/n-heptane kinetic model is presented that exhibits good agreement with the experimental data. The results also suggest that that the FIT instrument is a valuable tool for analysis of high pressure, low temperature chemistry and autoignition for future fuels in advanced combustion engines. 2. Methods 2.1 Fuel Ignition Tester. In this study, a Waukesha FIT was used to examine the reactivates of blends of n-butanol and n-heptane (an example future fuel). These blends were then compared to equal volumetric blends of Primary Reference Fuels (PRF), i.e. n-heptane and i-octane. The FIT is a constant volume combustion chamber device that allows a systematic determination of a DCN according to ASTM D717 (D717, 212). The ASTM D717 method uses the following correlation based on the measured ignition delay (ID) period after injection of a liquid fuel into the combustion chamber: = 171/ [ ] (4) The FIT operating temperature and pressure range, shown in Table 1, are ideal for examining the prominent low temperature combustion region of an HCCI engine as shown in the red shaded area in Figure 2(b). Table 1. Nominal operating parameters for the Waukesha FIT apparatus (ASTM D717). Volume.6 L Pressure 24 bar Wall Temperature 58 C (853 K) Equivalence Ratio

5 3 Cylinder Pressure [bar] BuOH_4v Avg = 22bar and 9K Heat Release Rate [J/deg] Deg ATDC Figure 2. (a) The Waukesha FIT apparatus for Derived Cetane Number (DCN) measurement and (b) a HCCI engine data pressure and heat release trace (from the present study) showing that the area of low temperature chemistry seen in an HCCI engine is in the same range as that tested in an FIT (6%v n-butanol / 4%v n-heptane at phi of.33, T in of 7 C, P in of.85bar, 15rpm and 16:1 compression ratio). 2.2 HCCI Engine. To demonstrate the performance of the proposed fuel blends in HCCI mode, a diesel engine was converted for use as an HCCI test bed. Specifically, one cylinder of a John Deere PowerTech 2.4L 424 turbo-diesel engine was modified to operate in HCCI mode. The existing in-cylinder fuel injector was disconnected in favor of using port fuel injection (via a gasoline-type injector ~2inches upstream of the intake valve) to produce a homogeneous mixture of air and fuel that is typical of HCCI operation. Additional modifications consisted of alterations to the intake and exhaust manifolds to allow isolation of the HCCI cylinder and the installation of an air preheater necessary to achieve the higher intake temperatures typically associated with HCCI operation. The piston head of the HCCI cylinder was also modified such that the compression ratio can be adjusted to allow HCCI tests at various compression ratios. The specifications of the HCCI engine are detailed in Table 2. Table 2. Homogeneous Charge Compression Ignition engine parameters for the experiments presented herein. Displacement Volume 69 cm 3 Compression Ratio 16:1 Speed 15 rpm Intake Temperature 7 C (343 K) Intake Pressure Bore Stroke.85 bar 86 mm 15 mm Fuel Equivalence Ratio Fuels and Chemical Kinetic Models. Future fuels with desired properties for optimal performance in these advanced combustion modes will most likely consist of mixtures of petroleum derived products (aromatics, straight and branched alkanes), alcohols, synthetic alkanes (straight and branched alkanes produced from natural gas, vegetable oil or animal fat), fatty acid methyl esters and natural gas (liquefied or compressed). Fuel blends such as these (with properties currently tailored for CI or SI engines) are beginning to see increased penetration into the global transportation fuel market. For example, the U.S. currently mandates a 1%v minimum ethanol blending requirement and the U.S. Environmental Protection Agency (EPA) recently reported that mixtures up to 15%v will be considered for future blends. 5

6 Butanol has received much attention in recent years as a bio-alternative to ethanol due to its higher energy density and its higher miscibility with other hydrocarbons. There have also been a wide array of kinetic studies of butanol (and its isomers) recently, and well-validated experimental data and chemical mechanisms exist (Dagaut, et al., 29) (Dagaut & Togbe, 28) (Saisirirat, et al., 211) (Zhang, 21) (Heufer, et al., 211) (Weber, et al., 211) (Sarathy, et al., 212) (Karwat, et al., 212) (Zhang, et al., 213). Recent research (Zhang, et al., 213) also suggests that n-butanol may possess unique chemical kinetic reactivity in that n-butanol (and blends with n-heptane) may be less reactive than n- heptane at low temperatures, but more reactive than n-heptane at high temperatures (i.e. engine combustion conditions). The present work seeks to validate this behavior and also to compare it to the known behavior of PRF s. Given this contrasting ignition behavior, the blend amounts of different fuels could be tailored to maximize HCCI performance. Table 3 shows the physical properties of the fuel mixtures used in this study. Viscosity, density, and bulk modulus measurements were performed by the authors while the RON values represent linear volumetric blending values based on publicly available numbers. Fuel n-butanol [vol%] Table 3. Fuel specifications of tested fuels n-heptane [vol%] i-octane [vol%] Viscosity [mm 2 /s] Density [g/cm 3 ] Bulk Modulus [kn/m 2 ] a nc nbuoh nbuoh nbuoh nbuoh nbuoh nbuoh b PRF PRF PRF PRF PRF PRF a Bulk modulus was calculated based on viscosity and speed of sound measurements. b n-butanol RON value from (Jin, et al., 211) (Aleiferis & van Romunde, 213). RON Accompanying the experimental data, zero-dimensional (sinlge-zone) models were run using the CHEMKIN and CHEMKIN-PRO chemical kinetic modeling packages. For comparison, several chemical kinetic mechanisms were used to examine the experimental results. For the n-butanol fuel blends, mechanisms developed by Saisirirat et al. (Saisirirat, et al., 211) and Sarathy et al. (Sarathy, et al., 212) were used. Several important distinctions exist between these two mechanisms, which are detailed in Table 4. Primarily it is important to note that while engine experiments were performed by Saisirirat et al., no engine modeling was done to attempt to match the data. Rather, the engine experiments were used to qualitatively interpret the mechanism. Therefore, the ignition behavior predicted by the Saisirirat et al. mechanism is based primarily on chemistry derived from Jet Stirred Reactor (JSR) data. Conversely, the Sarathy et al. mechanism has been validated against a very large data set, including the JSR data from Saisirirat et al. (211) as well as ignition experiments in both shock tubes and rapid compression machines. However, the Saisirirat et al. mechanism does include an n-heptane sub-mechanism, while the Sarathy et al. mechanism does not. Accordingly, for the present study, the authors have combined the n-heptane sub-mechanism from Mehl et al. (Mehl, et al., 211) with the Sarathy et al. (212) n-butanol mechanism, and further present some preliminary validation of this model via the experiments detailed in 3. Note that the Mehl et al. mechanism represents an update of the Curran et al. PRF model (Curran, et al., 22), which is the PRF sub-mechanism used in the Saisirirat et al. mechanism. 6

7 Table 4. Chemical kinetic mechanisms for n-butanol, n-heptane and iso-octane blends used in this study and the data sets against which these mechanisms have been validated. Mechanism Species Reactions Relevant Fuel Mixtures Saisirirat et /8 mol n- al. a BuOH/nC7 5/5 mol n- BuOH/nC7 1%v n-heptane 18%v nbuoh in nc7 37%v nbuoh in nc7 57%v nbuoh in nc7 7 Experimental Apparatus Jet Stirred Reactor HCCI Engine: 1 of 4 cyl diesel engine motored at 15rpm Sarathy et % nbuoh Laminar Flame al. b Measurements Temp Pressure Equivalence Range Range Ratios 53-17K 1atm.3,.5, 1 8 C intake temp n/a.3 353K 1atm % nbuoh Flat Flame n/a 4mbar 1.7 1% nbuoh Flat Flame n/a 15torr 1. 1% nbuoh Shock Tube K atm 1% nbuoh Shock Tube ,1. 16K 43atm 1% nbuoh Rapid Compression Machine K 15atm.5, 1., 2. 1% nbuoh Jet Stirred Reactor 53-17K 1atm.5, 1., 2. a Data for Saisirirat et al. from (Saisirirat, et al., 211) (Dagaut & Togbe, 21) (Dagaut & Togbe, 29) b Data for Sarathy et al. from (Sarathy, et al., 212) (Liu, et al., 211) (OBwald, et al., 211) (Hansen, et al., 211) (Heufer, et al., 211) (Stranic, et al., 212) (Weber, et al., 211) (Dagaut, et al., 29) 3. Results and Discussion 3.1 FIT Experimental Results. Fourteen different blends of iso-octane, n-heptane, and n-butanol were tested in the FIT, consisting of 28 test runs with 25 ignition measurements for each test run, totaling 35 individual tests. Figure 3(a) summarizes these test results, wherein n-heptane is the high reactivity fuel" and n-butanol or iso-octane are the low reactivity fuels. As noted above, the FIT records a Derived Cetane Number (DCN), which is approximately inversely proportional to RON (Ryan, 22) insofar as fuels with high RON exhibit low DCN and vice versa. It should also be noted here that the FIT (like the IQT) reports the ignition delay period based on an internal trigger response to the combustion pressure rising a set (small) amount above the base pressure. This technique can produce erroneous ignition delay periods since low temperature chemistry can result in a small pressure rise prior to primary ignition. For the high reactivity diesel fuels that these devices were originally designed to accommodate, the low temperature pressure rise is not problematic for DCN measurements since it occurs at time periods very close to that of the primary ignition event. However, for fuels with longer ignition delay periods such as those tested in the present study, the low temperature pressure rise can be misinterpreted as the primary ignition event. Accordingly, we report here both the FIT-reported ignition delay period along with an ignition delay period calculated from the peak pressure rise (PPR) as shown in Figure 3(a) and (b). The experimental results in Figure 3 support previous findings that fuel blends with similar Octane Number can exhibit very different ignition delay periods in an IQT or FIT. For example, a mixture of 4% n-heptane/6% n-butanol (nbuoh6 in Table 3) has a RON of 58, which is nearly identical to the RON of a PRF6 mixture by definition. However, the FIT-DCN of the former is nearly 2.5 times lower than that of the latter. The present experiments further show that n-butanol blends behaved unlike PRF blends when comparing the autoignition behavior as a function of the percentage of low reactivity component. Some of this behavior could be due to the tertiary carbon in iso-octane where the increased number of primary sites in iso-octane, on which the H-atom abstraction rate by OH is lower than secondary sites in n-alkanes. Similar results were seen with t-butanol in an IQT study by Hass et al. (Haas, et al., 211). In the Haas et al. study the concave down-like behavior of t-butanol/n-heptane blends was attributed to a lower OH hydrogen abstraction rate from t-butanol (compared to primary and secondary butanols), thus allowing more OH to be present to

8 attack the base n-heptane fuel. This behavior led to generally faster ignition times for t-butanol blends in comparison to the other 3 butanol isomers considered by Haas, et al. (211) n-heptane/n-butanol n-heptane/i-octane FIT-DCN FIT-DCN n-heptane/n-butanol n-heptane/i-octane %v Low Reactivity Fuel %v Low Reactivity Fuel Figure 3. Measured Derived Cetane Number for blends of n-heptane/n-butanol and n-heptane/iso-octane(a) per ASTM D717 and (b) measured from the peak pressure rise rate (PPR) as recorded by the FIT combustion vessel pressure transducer. 3.2 FIT Modeling Results. For the n-butanol/n-heptane blend simulations, the Sarathy (Sarathy, et al., 212) mechanism was updated with the n-heptane sub-mechanism from Mehl et al. (Mehl, et al., 211), which is a well-validated mechanism that has been used for modeling gasoline surrogate fuels such as PRF s. The Mehl mechanism was also used to simulate the PRF FIT experiments. Hereafter, the modified Sarathy et al. mechanism (including the n-heptane submechanism from Mehl et al.) will be referred to as the Sarathy* mechanism. Figure 4(a) and (b) are plots of measured and predicted FIT ignition delay period data for PRF fuels and n-butanol/n-heptane blends, respectively. 1 Figure 4. FIT Model results for (a) PRF and (b) n-butanol/n-heptane blends In Figure 4, the calculated ignition delay periods are compared against the two different types of ignition delay period measurements from the FIT experiments the corresponding calculated DCN values for these measurements can be found in Figure 3. The first measurement technique, which is that employed by the FIT DCN test method, defines the ignition delay period based on the point in which the measured pressure rises above a defined threshold. For fuels with long ignition delay periods that exhibit some level of low temperature reactivity (resulting in early pressure rise well before transition to a hot ignition event), this method can lead to false reporting of the ignition delay period. Accordingly, both the FIT-reported ignition delay period (blue line) along with an ignition delay period calculated from the peak pressure rise (PPR; red line) are reported here. Also plotted in Figure 4 (dashed red line) is the PPR ignition delay period, corrected for the vaporization time of the fuel. In its present configuration, the overall equivalence ratio of the FIT has an inherent margin of error (±.1 from the test average). Accordingly, the simulations were performed over a range of equivalence ratios and reported as the red and green shaded areas for the Saisirirat et al. and Sarathy* mechanisms, respectively. From these measurements the Sarathy* n-butanol mechanism clearly shows the better agreement with the experimental data. 8

9 3.3 HCCI Engine Experiments and Modeling. To further examine these fuels, HCCI engine experiments were also performed and simulated using the same chemical kinetic mechanisms. For this study, PRF4, nbuoh4, and nbuoh6 were chosen for comparison. The first two fuels were chosen because they were both found to ignite at near top dead center (TDC) for the engine configurations given in Table 2. The nbuoh6 blend was chosen to give a more extreme target for model comparison. Figure 5 is a plot of the measured instantaneous cylinder pressure and apparent rate of heat release for nbuoh4 and PRF4 fuels operating in the HCCI engine under the conditions listed in Table 2. Comparing this data to that in Figure 3 and Figure 4, it is apparent that under both engine and FIT conditions, the nbuoh4 fuel is found to ignite after the PRF4. This result is contrary to expectations given that nbuoh4 has a lower predicted RON than that of PRF4. It should be further noted that the difference in ignition delay is not as great in the HCCI engine as that observed in the FIT. This result can be attributed to the fact that the temperature and pressure in an engine are constantly rising from intake valve closure to top dead center; and so ignition differences, which may be large at low temperature, manifest themselves more subtly as the accumulated effects of differences as the temperatures and pressures are swept throughout the engine cycle. The effects of heat losses are also greater in the engine, so heat released as a result of chemical reactions do not lead to thermal runaway (i.e., ignition) at the same rate as in the FIT. For all CHEMKIN models, the in-cylinder temperature and pressure was slightly adjusted (+7 C and +2psi) to account for in-cylinder vs. intake conditions as well as to improve the agreement for both models. The initial conditions were then kept constant for all subsequent models to maintain consistency with the experimental tests. 3 Cylinder Pressure [bar] Heat Release Rate [J/deg] Deg ATDC Figure 5. HCCI engine data for nbuoh4 (solid lines) and PRF4 (dashed lines). Shown are cylinder pressure (thick lines) and net heat release rate (thin lines) for conditions listed in Table 2. Figure 6 through 8 are comparisons between the HCCI engine data and modeling results. The figures include the instantaneous in-cylinder pressure data (bar), along with the instantaneous apparent rate of heat release (J/deg) calculated from the in-cylinder pressure data. Figs. 6(b), 7(b) and 8(b) focus more closely on the low temperature component of the instantaneous apparent rate of heat release. For the HCCI conditions considered in this study, both the Sarathy* and Saisirirat et al. mechanisms are reasonably accurate in predicting the location of peak heat release (which corresponds closely to the location of CA5) as well as the point of peak pressure rise for the nbuoh4 data. However, as shown in Figure 7, the Mehl et al. mechanism (dotted line) compares well against the experimental data (solid line) and shows the PRF4 blend igniting earlier than the n-butanol blend. Conversely, the Saisirirat et al. mechanism predicts that this same PRF blend would ignite later than the n-butanol blend. Additionally, the Saisirirat et al. mechanism accurately predicts the crank angle at which the low temperature heat release occurs for both the nbuoh4 and nbuoh6 cases, but does not accurately predict the CA5 timing (Figure 6 and Figure 8). The deviations between the two models for the PRF fuels is likely a result of the Mehl et al. mechanism being essentially an updated version of the base PRF mechanism used in the Saisirirat et al. mechanism (i.e., the Curran et al. mechanism from 22 (Curran, et al., 22)). The slight retardation in the Saisirirat et al. model targeting the CA5 point for the nbuoh6 is likely due to the lack of validation of the mechanism against engine and ignition studies. In all cases, the zero-dimensional models over-predicted the magnitude of the heat release and the peak pressure. However, this result is expected with a zero-dimensional model since it does not take into consideration the heat transfer effects or non-homogeneities of equivalence or temperature gradients within the cylinder. Use of multi-zone modeling has proven to be very accurate at matching pressure and heat release profiles (Aceves, et al., 2) (Reaction Design, 9

10 21), but can be fairly computationally intensive and so it was decided to first examine the single-zone models and save the multi-zone modeling for the next phase of investigations on the present topic. Normalized Cylinder Pressure [bar] Actual nbuoh4 Sarathy* model Saisirirat et al. model Normalized Heat Release Rate [J/deg] Deg ATDC Deg ATDC Figure 6. HCCI data for nbuoh4 (solid line) vs. model data from Sarathy* (dotted line) and Saisirirat et al. (dashed line); (a) pressure and heat release, (b) low temperature heat release. Normalized Net Heat Release Rate [J/deg] Actual nbuoh4 Sarathy* model Saisirirat et al. model Normalized Cylinder Pressure [bar] Actual HCCI data Mehl model Saisirirat et al. model Normalized Heat Release Rate [J/deg] Deg ATDC Deg ATDC Figure 7. HCCI data for PRF4 (solid line) vs. model data from Mehl et al. (dotted line) and Saisirirat et al. (dashed line); (a) pressure and heat release, (b) low temperature heat release Normalized Net Heat Release Rate [J/deg] Actual HCCI data Mehl model Saisirirat et al. model Normalized Cylinder Pressure [bar] Actual HCCI data Sarathy* model Saisirirat et al. model Normalized Heat Release Rate [J/deg] Deg ATDC Deg ATDC Figure 8. HCCI data for nbuoh6 (solid line) vs. model data from Sarathy* (dotted line) and Saisirirat et al. (dashed line); (a) pressure and heat release, (b) low temperature heat release 1 Normalized Net Heat Release Rate [J/deg] Actual HCCI data Sarathy* model Saisirirat et al. model

11 3.4 Impact of Low Temperature Heat Release. In a recent study, (Rapp, et al., 212), Rapp and coworkers show that the normalized LTHR release is a very good prediction (linear correlation) of the compression ratio at which the CA5 point is at 6 ATDC. In that study, the LHTR is defined as the accumulated heat release occurring at temperatures < 1K, the high temperature heat release is defined as the heat release that occurs at temperatures > 1K and the normalized LTHR is defined as the ratio of the LTHR to the HTHR. In the present work, a fractional LTHR is defined as the fraction of the total heat release that occurs at low temperature (flthr). Specifically, we define the flthr as the integral of the low temperature heat release spike seen in the net heat release data (as shown more clearly in Figure 6(b) through Figure 8(b)) divided by the integral of the entire heat release trace. For all fuels tested, the LTHR was completed at a temperature of 925K and so this temperature was chosen as the cutoff for the flthr integral for all cases. In the study by Rapp and coworkers (212), the normalized LTHR was found to correlate with the compression ratio. In the present work, the measured flthr correlates well with the measured CA5 point. This correlation was found in both the HCCI engine test results and numerical modeling results using the Sarathy* mechanism (Figure 9). However, since experiments were conducted with only three fuel blends thus far, more data is needed to conclusively confirm that this correlation would be confirmed for all fuels at all HCCI operating conditions. It should also be noted that the Sarathy* model predicts that the relationship between flthr and CA5 is not precisely linear. However, the predicted flthr shown spans nbuoh1 to nbuoh6, which represents a wide range of fuel reactivity. The correlation between flthr and CA5 could potentially be considered linear over a narrower range of fuel reactivity. Lastly, as nbuoh6 did not have a clearly defined LTHR (since it occurs very closely to the HTHR), it becomes much more difficult to determine the LTHR for fuels that react in this way; although defining a cutoff temperature (such as the 925K used herein) does give an value for flthr. The model predictions for CA5 as a function of flthr from the Sarathy* mechanism in Figure 9 were shown to vary quadratically with an R 2 value of.996. The reasons for this quadratic variation are discussed below. CA5 [deg ATDC] nbuoh4 PRF4 nbuoh6 Sarathy* Model Fractional LTHR Figure 9. Predicted and measured variation in CA5 as a function of fractional low temperature heat release (FLTHR) for HCCI engine conditions of Table 2. Experimental results are shown for 3 different fuel blends. Model results are for nbuoh1 to nbuoh6 blends. 3.4 Relation between flthr and CA5. A first attempt at an analytical understanding of the relationship between flthr and CA5 is presented in this section. Figure 1(a) is a plot of temperature history for the ignition of two fuel blends (nbuoh6 and pure nbuoh) at constant volume for 24bar initial pressure and 82K initial temperature. It is apparent that the nbuoh6 exhibits LTHR, whereas the neat nbuoh exhibits no LTHR at these conditions. The net effect of the LTHR for the higher reactivity fuel blend is to raise the temperature of the homogeneous mixture, which results in a reduced ignition delay period before the primary ignition event. 11

12 To illustrate how increased flthr results in an advance in CA5 in an HCCI engine, one can begin by assuming that the overall ignition delay of a premixed homogenous fuel/air varies with temperature and pressure according to the Arrhenius form: = exp (5) where P is the pressure, T the temperature (which, in an HCCI engine varies with crank angle and therefore time), and A, B, and n are constants. Equation 5 is often presented in the form of an Arrhenius-type plot as shown in Figure 1(b) for nbuoh6 at 4 bar and φ=.33. If one assumes that the ignition delay is only weakly a function of pressure (Livengood & Wu, 1955) (Douaud & Eyzat, 1979), then Equation 5 can be simplified as exp (6) Temperature [K] 2 nbuoh1 nbuoh Time [ms] ln(ignition delay [ms]) ln T Temperature [K] Figure 1. Ignition plots for nbuoh6 showing (a) effect of LTHR on shortening ignition delay vs. pure nbuoh and (b) the Arrhenius-type relation between ln(ignition delay) and temperature for nbuoh6 simulated with Sarathy* at 4bar and φ=.33. In an HCCI engine with fuel/air mixtures with ignition behavior that follows Equation 6, an Arrhenius plots such as Figure 1(b) would be valid over some small pressure range such as for several crank angles near top dead center (TDC) in a motored engine cycle. Additionally, Figure 1(b) is nearly linear for temperatures greater than the negative temperature coefficient (NTC) region, i.e. greater than about 9K for most hydrocarbons. Assuming a linear relation between temperature and ln(ignition delay) allows the derivative of Equation 6 to be simplified as: ln (7) where C is a constant equal to the slope of the high temperature section of Figure 1(b). Equation 7 can be rearranged to yield the following relation: ln ln exp Equation 8 can be put into the context of an HCCI engine by recognizing that the ignition delays, τ 1 and τ 2 are directly related to crank angles at which the fuel mixture within the cylinder combusts. From Figure 1(a) one can see that the time for the mixture to reach full heat release is essentially the same time as for the mixture to reach the 5% heat release, or in other words the τ can be thought of as the CA5 point for different fuels. The, can be thought of as the flthr that shortens the ignition delay relative to the pure component fuel with no LTHR. Furthermore, the flthr can be related to the overall heat release via the following: (8) 12

13 (9) where CA L is the crank angle at which low temperature heat release is completed, CA o is some initial reference crank angle, T o some initial reference temperature, T l the temperature at which the flthr is completed, and T f the temperature upon complete heat release and is essentially the maximum temperature seen within the engine. Combining Equations 8 and 9, the final general form relating flthr to CA5 is found to be 5 (1) where τ LR is the CA5 time of the base, low-reactivity, fuel and C Q is a constant combining the total heat release of the fuel and the slope of the Arrhenius plot. Finally, since flthr is a small number, a simplified Taylor series expansion of Equation 1 results in the following: ! + 2! ! Equation 11 is consistent with the results of Fig. 9, which showed that numerical predictions of CA5 varied quadratically with flthr. Since the low temperature heat release occurs at lower pressure it is reasonable to examine the flthr as measured at low pressure in the FIT (flthr FIT ) to see how well it relates to that seen in the engine. To first examine the correlation between the low pressure ignition delay data and the HCCI engine data, the Sarathy* mechanism is again used since model results for FIT and engine conditions can be easily obtained for all fuels. Figure 11(a) is a plot of the predicted CA5 for HCCI engine model using the Sarathy* mechanism against the fractional low temperature heat release predictions by the same mechanism under FIT conditions (flthr 24bar-model ). A 2 nd order polynomial in the form of Equation 11 was fit to the data in Figure 11(a) with an R 2 of.998. Sarathy* Model CA bar Sarathy* FIT model flthr Ratio: FIT Ig Dly / dpmax Figure 11. Simple model fits relating FIT to the observed HCCI engine data (a) 24bar FIT model vs. model predictions for CA5 and (b) actual ratio of FIT ignition delay to ignition delay found from max dp/dt curve. 24bar Sarathy* FIT model flthr (11) In its current configuration, measuring flthr directly from the FIT data is difficult because the sample rate at which the pressure data is recorded by the FIT is not sufficient to accurately resolve the LTHR. However, a proxy for this method is proposed. As shown in Figure 4, the FIT-reported ignition delay can be much earlier than the actual ignition delay (as defined by maximum pressure rise). As explained above, the FIT measures ignition delay as the time difference between injection and the chamber pressure rising above a set trigger point. If one makes the simplifying assumption that the FIT reported ignition delay is proportional to the LTHR then the ratio of FIT reported ignition delay to the ignition delay found from dpmax (of FIT pressure trace) should have a linear correlation against the flthr this is indeed the case as is shown in Figure 11(b), which as an R 2 of.941. Also, the error bars for the data in Figure 11 are large due to the compounding effect of a ratio of small errors (see Figure 4 for the magnitude of the individual FIT errors, which are quite small) this same effect is also seen in Figure 12, which uses the trends found in Figure 11 to predict a simple model of CA5 based on the FIT data; note that an excellent correlation is found with experimental data.

14 2 15 CA5 [deg ATDC] Ratio: FIT Ig Dly / dpmax Figure 12. Simple model showing the relation between the ratio of FIT ignition delay to ignition delay found from max dp/dt curve vs. the CA5 found in an HCCI engine. One caveat to the proposed relationships in Figure 12 is that they are based on the ignition delay data for a n- heptane/n-butanol blend. A natural outcome of the derivation of Equation 11 is that only fuels with similar ignition delay behaviors can be regressed using the same general equations. While there will certainly be a flthr:ca5 relationship for every fuel (or fuel blend), the exact equation will depend on the blended fuels such that a given fuel blend may have a different relation (i.e. the curve in Figure 9 will shift based on the fuels in the blend) but it is possible that these relations may be assumed to be rooted in the most and least reactive fuels in the blend. In other words, two different fuel blends may fall on the same blend line (i.e. Figure 9) if the two fuel blends share the same least and most reactive fuels as these fuels will control the extreme bounds of reactivity. 4. Conclusions In this study two sets of fuel blends were tested in varying amounts of blend ratios in both a FIT and an HCCI engine. Two chemical models were also examined and tested against the experimental data; the Sarathy et al. mechanism updated to include a current n-heptane sub-mechanism by Mehl et al. (dubbed Sarathy*) showed the best ability to model the results obtained from both sets of experiments. The FIT seemed to accurately predict relative fuel ignition rankings, though the magnitude of the differences was greater than that seen in an HCCI engine. However, the FIT experimental data proved quite useful, serving as further validation of the examined models at low temperatures. Lastly, the fractional LTHR showed great promise of being an indicator of CA5 location based on the derivation and data comparison herein. Investigating the influence of flthr and the possibility of obtaining this information from an FIT will be the topic of future research. Acknowledgements This research was funded in part by the Colorado Center for Biofuels and Biorefining (C2B2). The authors would also like to thank Harrison Bucy and Nathan Pekoc for their technical assistance with the FIT as well as Aaron Gaylord, Michael Herder, Scott Salisbury, Ryan Farah, Ryan Peters, and Nathan Zeleski for their assistance in converting the John Deere 424T engine to operate in HCCI mode. Co-author SMS acknowledges funding from the Clean Combustion Research Center (CCRC) at KAUST, and from Saudi Aramco under the FUELCOM research program. References Aceves, S. et al., 2. A multi-zone model for prediction of HCCI combustion and emissions. s.l.:sae

15 Aleiferis, P. & van Romunde, Z., 213. An analysis of spray development with iso-octane, n-pentane, gasoline, ethanol and n-butanol from a multi-hole injector under hot fuel conditions. s.l.:fuel. Bessonette, P. et al., 27. Effects of fuel property changes on heavy-duty HCCI combustion. s.l.:sae Bezaire, N., Wadumesthrige, K., Ng, K. & Salley, S., 21. Limitations of the use of cetane index for alternative compression ignition engine fuels. s.l.:fuel. Bogin, G. et al., 211. Numerical and Experimental Investigation of n-heptane Autoignition in the Ignition Quality Tester (IQT). s.l.:energy and Fuels. Bogin, G. et al., 213. Ignition Quality Tester (IQT) Investigation of the Negative Temperature Coefficient Region of Alkane Autoignition. s.l.:dx.doi.org/1.121/ef31738b, Energy&Fuels. Ciatti, S., 212. The Gasoline Diesel: Research at a National Lab aims to break barriers in fuel economy and emissions performance. s.l.:mechanical Engineering Magazine. Curran, H., Gaffuri, P., Pitz, W. & Westbrook, C., 22. A Comprehensive Modeling Study of iso-octane Oxidation. s.l.:combustion and Flame. D689, 212. Standard Test Method for Determination of Ignition Delay and Derived Cetane Number (DCN) of Diesel Fuel Oils by Combustion in a Constant Volume Chamber. s.l.:astm International: West Conshohocken, PA. D717, 212. Standard Test Method for Determination of Derived Cetane Number (DCN) of Diesel Fuel Oils Fixed Range Injection Period, Constant Volume Combustion Chamber Method. s.l.:astm International: West Conshohocken, PA. Dagaut, P., Sarathy, S. & Thomson, M., 29. A chemical kinetic study of n-butanol oxidation at elevated pressure in a jet stirred reactor. s.l.:proceedings of the Combustion Institute. Dagaut, P. & Togbe, C., 28. Oxidation kinetics of butanol-gasoline surrogate mixtures in a jet-stirred reactor: Experimental and modeling study. s.l.:fuel. Dagaut, P. & Togbe, C., 29. Experimental and Modeling Study of the Kinetics of Oxidation of Butanol-n-Heptane Mixtures in a Jetstirred Reactor. s.l.:energy and Fuels. Dagaut, P. & Togbe, C., 21. Experimental and modeling study of the kinetics of oxidation of ethanol-n-heptane mixtures in a jetstirred reactor. s.l.:fuel. Dec, J. E., 29. Advanced compression-igniton engines understanding the in-cylinder processes. s.l.:proc. Combust. Inst.. DOE, U., April, 21. Homogeneous Charge Compression Ignition (HCCI) Technology: A Report to the U.S. Congress. s.l.:u.s. Department of Energy, EERE, OTT. Douaud, A. & Eyzat, P., Four-Octane-Number Method for Predicting the Anti-Knock Behavior of Fuels and Engines. s.l.:sae 788. EPA, 212. Inventory of U.S. Greenhouse Gas Emissions and Sinks: s.l.:u.s. Environmental Protection Agency, EPA 43-R Haas, F., Ramcharan, A. & Dryer, F., 211. Relative Reactivities of the isomeric Butanols and Ethanol in an Ignition Quality Tester. s.l.:energy and Fuels. Hansen, N., Harper, M. & Green, W., 211. High-temperature oxidation chemistry of n-butanol - experiments in low-pressure premixed flames and detailed kinetic modeling. s.l.:phys. Chem. Chem. Phys.. Heufer, K. et al., 211. Shock tube investigations of ignition delays of n-butanol at elevated presures between 77 and 125 K. s.l.:proceedings of the Combustion Institute. Hirsch, R. L., Bezdek, R. & Wendling, R., 25. Peaking of World Oil Production: Impacts, Mitigation and Risk Management. s.l.:us Dept. Energy/National Energy Technology Laboratory. Jin, C. et al., 211. Progress in the production and application of n-butanol as a biofuel. s.l.:renewable and Sustainable Energy Reviews. 15