A study on auto-ignition of poly(oxymethylene) dimethyl ethers and their mixtures with the primary reference fuel 90

Abstract

Presently, the transportation sector is struggling to reduce its share of fossil fuels, by employing renewable fuels which are carbon-neutral and, in addition, may reduce engine-out emissions of soot and particulate matter. Among the renewables, poly(oxymethylene) dimethyl ethers (OMEn, n = 1-5; collectively named as OMEs) have an excellent soot reduction potential and can act as a drop-in fuel component in conventional engines due to their high cetane numbers and fast evaporation rates. A comprehensive understanding of the fundamental combustion properties of OMEs, such as ignition delay times (IDTs) and laminar burning velocities (LBVs), is essential for the evaluation of their engine application potential and the development of safer and more fuel-efficient engines (LBVs). In this work, IDTs of stoichiometric mixtures of dimethyl ether (OME0), OME1, OME2, iso-OME2 (trimethyl orthoformate, i.e., HC(OCH3)3), and OME4 with synthetic air diluted 1:5 with nitrogen were measured behind reflected shock waves in a shock tube at T = 800-2000 K for atmospheric (1 bar) and elevated pressures at 4 and 16 bar. In addition, since OMEs are discussed as suitable alternative blending compounds for fossil-based fuels, the effect of the addition of OME1, OME2, and iso-OME2 to a gasoline surrogate, the primary reference fuel 90 (PRF90: 90% iso-octane + 10% n-heptane by liquid vol.), on IDTs was investigated. In detail, IDTs of mixtures of PRF90 / synthetic air and of blends (by liquid vol.) of 70%OME1 + 30% PRF90, 70%OME2 + 30% PRF90, and 70% iso-OME2 + 30% PRF90 with synthetic air, all diluted 1:5 with nitrogen at stoichiometric condition, were measured in a shock tube in the temperature range T = 950-2000 K for pressures between 1-16 bar. The experimentally determined IDT data sets have been compared with the results of predictions made using the in-house reaction DLR-Concise model by Kathrotia et al. [1] and public domain reaction models taken from the literature. Furthermore, the data obtained for IDTs of the neat and blended fuels are supplemented with corresponding experimental data for the laminar burning velocities (LBVs) published by Ngugi et al. [2-6]. These measurements were performed using the cone angle method at p / bar = 1, 3, and 6, fuel-air ratios (φ) ranging between 0.6 and 1.8, and at a constant preheat temperature of 473 K. The results obtained are augmenting the data sets for evaluating the performance of the used reaction models. The comparison of the experimental data obtained under similar conditions for the IDTs - as well as for the LBVs of pure OMEs (OME0, OME1, OME2, and OME4)-are made to bring out the effect of chain length on the reactivity of OMEs. The measured values for IDTs of the four OMEs converge at temperatures above 1450 K independent of pressure, whereas at temperatures below 1450 K, the measured IDTs are shortest for OME4 and longest for OME0 (DME). From this observation, it is concluded that the reactivity of OMEs increases with an increase in the chain length. This finding is supported by the results of laminar burning velocity measurements, which are highest for OME4 at all pressures and over the entire equivalence ratio range considered. Further, the IDT data for OME2 and OME4 are close for all the conditions investigated indicating that for OMEs, the increase in reactivity is reducing as chain length increases. Similar to this, LBV values of OME2 and OME4 are close for φ ≤ 1.0. The comparison between measurements and predictions using DLR-Concise [1], Cai et al. [7], and Niu et al. [8] models reveal that the three models satisfactorily predict the measured IDTs of pure OMEs for most of the conditions. On the other hand, larger deviations were observed between measured and calculated laminar flame speeds (LFSs) for most of the OME-air mixtures and conditions covered. The measured IDT data revealed that OME2 and OME4 exhibit a pre-ignition behavior at T ≤ 1100 K, particularly at 4 and 16 bar, as demonstrated by an earlier increase in OH* and CH* before the main ignition. A strong perturbation on the pressure profile due to pre-ignition heat release was also observed. The comparison between measurements and predictions using the DLR-Concise [1] as well as the models of Cai et al. [7] and Niu et al. [8] indicates that the models satisfactorily predict the main ignitions within experimental uncertainty. Further, the models of Cai et al. [7] and Niu et al. [8] adequately account for the pre-ignition behavior observed in the measurements. The results show that pre-ignition is a consequence of the reaction behavior at low temperatures. Since the low-temperature chemistry is absent in the DLR-Concise mechanism, the modeling results do not show pre-ignition. The comparison of the measured data for iso-OME2 and OME2 shows that the two fuels have similar IDTs. Similar to the IDTs, the measured LBVs of iso-OME2 and OME2 are relatively similar in the fuel-lean up to the stoichiometric domain. However, under fuel-rich conditions, the LBVs of OME2 are significantly higher, i.e., by up to 30% at fuel-air ratio φ > 1.50 and 1 bar. For all pressures, the DLR-Concise model matches the measured ignition delay times data of iso-OME2 for T ≥ 1250 K, but overpredicts the measured LBVs in the whole stoichiometry regime. The results obtained for the blended fuels are compared to those of the pure fuels (OME1, OME2, and iso-OME2) and PRF90 for the same conditions. The results show that IDTs of the fuel blends (OME1 / PRF90, OME2 / PRF90, and iso-OME2 / PRF90) are shorter than those of PRF90 and longer than those of the pure OMEs, showing that the addition of OMEs increases the reactivity of PRF90 since the reactivity of pure OMEs is significantly higher than that of PRF90. This finding is also demonstrated by an increase in the LBVs of the fuel blends [2-6]. The impact of increasing the OME1 fraction from 0-100% on the IDTs of OME1 / PRF90 blends is inferred from measurements as well as from predictions with the DLR-Concise model. The results show that IDTs of the blends decrease in a weakly non-linear fashion by increasing OME1 fractions from 0-50%. The reduction of IDTs of the blend is stronger for blends with over 50%OME1 fractions. The comparison of measured and predicted data showed that the DLR-Concise model satisfactorily reproduced the experimental data for IDTs and LBVs of the blended and the neat fuels within the experimental uncertainty. In the current study, significant new data for ignition delay time data of pure OMEs (OME0, OME1, OME2, iso-OME2, and OME4) and of blends of OME1, OME2, and iso-OME2 with a gasoline surrogate (PRF90) in the mid- to the high-temperature regime (T = 800-2000 K) at atmospheric (1 bar) and elevated pressures at 4 and 16 bar were obtained. In particular, this work characterizes pre-ignition behavior, which was observed in the shock tube experiments in the low-temperature regime. The results make it possible to test the implementation of low-temperature chemistry of OME2 and OME4 in the chemical kinetic reaction models. The results of this systematic analysis of the five neat fuels and the blended fuels under consideration have broadened the experimental data sets in terms of the chosen experimental conditions (pressure, temperature, and fuel-to-air ratio) required for rigorous testing, thus improving chemical kinetic models focusing on fundamental combustion properties for OMEs.