Baumgarten

6.4 Homogeneous Charge Compression Ignition (HCCI) 255

Fig. 6.25. Comparison of cylinder pressure histories. a SI operation (gasoline), b HCCI operation (labeled ATAC) with methanol [37]

Fig. 6.26. Dual-mode operation

Najt and Foster [57] investigated the HCCI-concept for four-stroke-engines for the first time. The authors used a machine with variable compression ratio and heating of the intake air, and the fuels were mainly blends of heptane and octane. It was shown that high compression ratios allow a decrease in intake air temperature but result in excessive heat release and knocking. Thring [91] used a fourstroke engine with variable compression ratio in order to investigate the behavior of gasoline and diesel HCCI combustion. As a consequence of the limited operating range he suggested a hybrid combination of HCCI (part load) and conventional combustion (high and full load). Further basic investigations have been performed by Stockinger et al. [85] and Ayoma et al. [7].

Altogether, these early investigations have already shown that an important reduction of NOx and soot raw emissions could be achieved, while in most cases the CO and HC emissions increased because of the low combustion temperatures. Furthermore, it has turned out that HCCI combustion is only possible at moderate loads and engine speeds, Fig. 6.26, due to uncontrollable and premature knocklike combustion at higher loads and partial burning at very low loads or high engine speed (Sects. 6.4.4 and 6.4.5). In the case of higher loads and especially at full load, conventional diesel spray combustion or gasoline SI-combustion have to be used. This so-called dual-mode concept is only attractive for applications with a high degree of part load. However, it is expected that the further development of

6.4 Homogeneous Charge Compression Ignition (HCCI) 257

The major difference between the conventional diesel combustion and the HCCI process lies in the composition of fuel vapor at ignition. Because of its very heterogeneous fuel-air mixture, conventional combustion starts at those local mixture regions having the best ignition conditions, and then the heat generated from the combustion of these local sections enhances ignition of the neighboring regions. After ignition, the conventional diesel combustion begins with a very fast high temperature oxidation of the premixed air-fuel quantity, which has been generated during ignition delay (premixed peak). The rate of heat released by the subsequent diffusion burning depends on the velocity of the turbulent mixing of unburned or partially burnt fuel and air. In Fig 6.27, the premixed peak near TDC and the subsequent comparatively slow diffusion burning are shown.

In the case of HCCI combustion on the other hand, the whole cylinder charge reacts simultaneously. The whole charge is firstly oxidized according to the LTO chemistry and then, after a second delay (NTC region), passes to the high temperature oxidation.

In order to achieve a homogeneous air-fuel mixture, the time available for mixing must be maximized, and fuel injection must occur at early crank angles. However, this often leads to an early start of LTO and HTO reactions. Because the whole charge reacts simultaneously, the heat is released in a very short time, resulting in a considerable pressure rise, increase of noise (dp/dΜ) and peak pressure compared to the conventional diesel combustion.

Here, three main challenges of HCCI combustion, the formation of a homogeneous air-fuel mixture, and the prevention of premature ignition timing and the prevention of excessive heat release rates, which limit the use of HCCI combustion to part load today, become already apparent.

The theoretical and practical roots of HCCI-combustion concepts are ultimately credited to the Russian scientist Nikolai Semenov, who began pioneering work in the field of ignition in the 1930s. Two relevant aspects of his work are the thermal and the chemical ignition theories [44, 63, 64]. The thermal ignition theory postulates that the combustion process can be initiated only, if sufficient molecular collisions are occurring, i.e. sufficient temperature and pressure conditions must be existent. The chemical or chain theory of ignition hypothesizes, that combustion involves a process of branching chemical reaction chains, initiated at active chemical centers. Spontaneous ignition occurs if the number of chains being initiated exceeds the number of chains being terminated. In the 1970s, skeletal kinetic models, based on degenerate-branched-chain and class chemistry concepts, were developed for the prediction of auto-ignition delay time in engines [32] and formed the basis for the development of further extended chemical kinetic models. In the following, the most important chemical reactions during LTO and HTO of alkanes, which are used in the chemical kinetic models, are explained.

Low Temperature Oxidation of Alkanes

The time of occurrence and the following heat release of the first stage ignition depend on the fuel molecular size and structure. The LTO reactions generally oc-

R, R' and R'' are fuel molecules and “ • ” denotes fuel radicals according to Warnatz et al. [97]. The OH• radicals produced by these reactions oxidize the hydrocarbons (reaction 6.2),

and the increasing reaction rate of these exothermic reactions (reactions 6.3 to 6.12) result in the first heat release (LTO). However, backward reactions become dominant as temperature increases due to heat release and further compression. In order to show that the reactions in both directions are important (especially in Eqs. 6.3 and 6.5), the “ ” symbol is used. With increasing temperature, the formation of oxidizer (OH• ) degenerates, leading to the degeneration of the first-stage heat release. This mechanism is called degenerate chain branching. It is responsible for the increase of ignition delay with rising temperature (Negative Temperature Coefficient (NTC), Fig. 6.28) and explains the existence of the NTC region between LTO and HTO during the two-stage HCCI combustion.

NTC Region

At temperatures between approx. 800 K and 1000 K, the fuel radicals from reaction 6.2 feed the following two reactions,

and result in an accumulation of H2O2, which remains relatively inert as long as the temperature is below about approx. 1000 K [31].

Temperature increases due to compression, and above 900-1000 K the chain

Fig. 6.28. Negative temperature coefficient for hydrocarbon ignition. a [84], b [103]

6.4 Homogeneous Charge Compression Ignition (HCCI) 261

Extended Chemical Models

The chemical kinetics of auto-ignition and combustion is extremely complex. In the 1970s the first reduced kinetic models were developed for prediction of autoignition delay time in an engine [32]. Since the 1980s, more and more detailed chemical kinetic models have been published. However, there are hundreds of species and reactions that have to be modeled, even if simple hydrocarbons like butane [29] are used. In the case of fuels with higher molecular weight such as n- heptane [13], the modeling becomes even more complex. Despite the complexity, several approaches using detailed chemical mechanisms have been reported. Some authors use single-zone models [3, 31, 42], others multi-zone ones along with CFD [2, 45]. For example, Groenendijk et al. [31] used a detailed single-zone model (857 species and 3606 reactions for iso-octane and 544 species and 2446 reactions for n-heptane). These models are very useful for fundamental analyses of the effect of fuel or EGR composition on ignition timing and combustion.

Nevertheless, besides a detailed description of the relevant chemistry, an accurate consideration of three-dimensional turbulence and inhomogeneities inside the combustion chamber is also of fundamental importance for the HCCI combustion process. However, the combination of CFD and detailed chemistry results in very time-expensive and often impractical calculations. From this point of view, a combination of reduced chemical models and CFD seems to be the best solution. The reduced chemical models must be able to map the important reactions required to calculate the main features like ignition timing, heat release, temperature, and pressure histories and fuel consumption. Li et al. [51] have developed a reduced kinetic model for primary reference fuels (PRF). Zheng et al. [106] used this model in order to simulate HCCI first-stage ignition. The model includes 29 reactions and 20 active species. The simulations of Zheng et al. [106] agree well with experimental data concerning pressure, ignition timing, and first-stage ignition heat release. Zheng et al. [107] successfully extended this model through the entire HCCI combustion process. The new model includes 69 reactions and 45 active species and combines the chemistry of the low, intermediate, and the high temperature regions.

6.4.3 Emission Behavior

As already described, one of the most important benefits of HCCI combustion is the enormous reduction of NOx raw emissions of 90–98% in comparison to conventional combustion [17, 24, 43, 61]. As far as homogeneity is concerned, a large degree of mixture inhomogeneity can be tolerated without resulting in increased NOx formation [28]. The biggest part of the very low total NOx emissions is due to the NO formation mechanism from N2O [16, 53, 97]. This mechanism is important for combustion processes with high air excess ratios, low temperatures, and high pressures [73].

Today, HCCI combustion is limited to part load operation, where the mixture equivalence ratios and thus also the combustion temperatures of the homogene-

6.4 Homogeneous Charge Compression Ignition (HCCI) 263

Another important benefit in the case of HCCI diesel combustion is the possibility of reducing NOx and particulate matter simultaneously [17, 24, 43, 65]. The formation of soot requires fuel-rich zones (Ι> 1.7) and temperatures above 1400 K [8, 73], which are not present due to the homogeneous lean mixture. However, fuel deposition on the walls may result in poor mixture preparation and in local fuel-rich regions that are subject to incomplete combustion and produce soot. This can especially happen during early direct injection, because pressure and temperature inside the cylinder are low and fuel penetration is increased.

Compared to conventional combustion concepts, HCCI combustion usually results in significantly higher HC emissions [61, 67, 73, 87]. The HCCI diesel experiments of Schlotz [73] for example showed approximately five times more HC emissions in comparison to the conventional diesel combustion. This is caused by the low combustion temperatures due to the lean mixtures and/or the high EGR rates. EGR is usually needed to lengthen the combustion duration in order to avoid extreme heat release rates. In the case of excessive EGR this can result in misfiring and thus in a significant increase of HC emissions. On the other hand, EGR can also help to reduce the amount of unburned hydrocarbons, because they get a second chance to take part in the combustion process.

In the case of early direct injection, fuel deposition on the cylinder walls and especially in the top-land may also result in an excessive increase of HC emissions. In this case, the fuel injection system must be specially adapted to combustion chamber geometry and gas density, Sect. 6.4.4. The lower the volatility of a fuel, the more serious the problem of wall wetting.

The reduced combustion temperatures are responsible for partial burning and for decreased post-combustion oxidation rates within the cylinder, especially near the walls. For this reason, HCCI combustion also typically results in higher CO emissions than conventional diesel or spark-ignited combustion. The HCCI diesel experiments of Schlotz [73] show approximately ten to twenty times more CO emissions compared to the conventional diesel combustion.

The amount of HC and CO emissions directly influences fuel consumption, because both components contain chemical energy that has not been released during combustion. If a proper phasing of combustion relative to the engine cycle is realized, the HCCI process can approximate the ideal Otto cycle (combustion at constant volume) because of the high heat release rates and the very short combustion duration. If there is no partial burning or misfiring, this generally results in high efficiency [20]. Furthermore, the low combustion temperatures result in a reduced loss of heat to the engine. In the case of the gasoline engine, an additional advantage in efficiency is provided by the omission of throttle losses. The theoretical potential of reducing fuel consumption at partial load is comparable to that of the spray-guided direct injection technique [94, 23]. HCCI fuel efficiencies comparable to those of conventional diesel combustion at part load have been reported by several authors [7, 34]. However, if fuels with low volatility and high ignitability like diesel are used, this beneficial effect of HCCI combustion cannot always be realized because of insufficient mixture preparation, fuel wall impingement, partial burning, and poor combustion phasing (e.g. premature ignition).