Development of a method to investigate the influence of engine oil and its additives on combustion anomalies in hydrogen engines

In a hydrogen engine, various influencing factors can lead to pre-ignition. Research activities on the mechanisms of pre-ignition have already been conducted in turbocharged gasoline engines, where potential causes for pre-ignition have been described. Optical studies on a full-scale gasoline passenger car engine revealed that ignition occurred at different locations due to randomly drifting droplets (oil, fuel). While ignition at glowing particles could not be ruled out, the most plausible theory emerged as the increased detachment of oil droplets from the cylinder wall film [3, 9].

Diverse studies involving (PFI) and (DI) have already identified several potential initiators for combustion anomalies in hydrogen engines. Aside from the ignition system, previous experiments have illustrated the impact of oil temperature and deposits resulting from the oil [2, 9]. In Fig. 2, alongside the various causes of pre-ignition, the potential sources of engine oil introduction into the combustion chamber are illustrated. Engine oil can enter the intake air through the crankcase ventilation, as well as through the bearings of the turbocharger and the valve stem guides. On one hand, the entry of oil into the combustion chamber can lead to oil deposits [10] and on the other hand, there is a possibility that oil droplets may enter the combustion chamber, potentially causing pre-ignition. This formation hypothesis by (LOI) is shown schematically in Fig. 3.

If these oil droplets are in the temperature range between \(200 \,\,and \,\, 300\,^{\circ } \hbox {C}\) [3], evaporization (a) (Fig. 3) of the more volatile components begins, so that with increasing temperature at the droplet surface, an oil–air mixture (\(\lambda _\text {oil}\)) gradually forms. Upon reaching the self-ignition conditions of the oil (local temperature and \(\lambda _\text {oil}\)), oxidation (b) (Fig. 3) and flame propagation occur. If the flame front is not quenched prematurely, the surrounding hydrogen–air mixture (\(\lambda _\mathrm {H_2,\,local}\)) is ignited by the released energy from the oil droplet. However, it is also possible that the oil droplet may change completely to the gas phase (droplet lifetime) before the self-ignition temperature is reached, due to its boiling behaviour, size and/or cycle time .

Multiple sources and mechanisms contribute to the loss of engine oil in an internal combustion engine. In Fig. 4, the potential sources of oil consumption in a turbocharged internal combustion engine are schematically depicted. These sources are divided into leakage losses directly escaping from the engine housing, oil losses through valve stem seals and the sealing of the turbocharger lubrication, oil transport through crankcase ventilation due to blow-by and oil consumption off the piston system, consisting of the piston, piston rings, and liner. The underlying mechanisms depend on a variety of engine-specific conditions such as engine design and configuration, load point, or wear condition. Therefore, the precise contribution of each source to oil consumption and the resulting absolute amount cannot be universally stated.

However, it can be reasonably assumed that the valve stem seals of modern engines and the direct escape through lubricant leaks, in a proper and well-maintained condition, contribute negligibly to the total oil consumption. Accordingly, the oil consumption shares are primarily caused by the lubricant supply of the piston system, as well as crankcase ventilation and sealing of the turbocharger lubrication to the exhaust or charge air side.

For a better understanding about the percentage of the individual oil consumption sources, Fig. 5 shows measurement results for three representative steady-state load points (urban, intercity, motorway) based on a passenger car diesel engine. The average load demand of the analysed load points increases in the order shown. The average intercity point has a comparable torque in relation to the urban point, but at a higher engine speed level. In addition to the percentage breakdown of the sources, the percentage change in the overall consumption and the qualitative increase or decrease (arrows) in the individual quantities between the load points are also plotted. The evaluation emphasises that the overall lube oil consumption (LOC) increases with increasing load demand, whereby the main proportion comes from the piston–ring-liner system. The blow-by oil increases at higher engine loads and speeds due to the enhanced blow-by mass flow. The proportion of lube oil consumption caused by the turbocharger lubrication is highest in the low-load range. A possible explanation for this is the unfavourable pressure conditions in the turbocharger at low-load operation.

The experimental testing was conducted on two different test rigs. First, a 2-L four-cylinder gasoline engine was used as a powertrain for passenger cars (PCs) and light commercial vehicle (LCV). Second, a 12.8-L 6-cylinder (HD) engine was employed. Both engines were modified for hydrogen operation through adjustments to the charging system, injection, and ignition system [19, 20] (Tables 2, 3).

A significant difference between the two test engines is the mixture formation. The PC engine features internal mixture formation, while the HD engine has external mixture formation. In addition, for both engines, a dosing lance was integrated into the intake manifold at cylinder 4 (Fig. 8) to introduce oil droplets as close to the valve as possible, minimizing the likelihood of wall deposits.

The objective was to select an operating point where the hydrogen engine runs stably with minimal combustion anomalies, and the compression end temperature is within the range of oil autoignition temperature. In addition, the time for the evaporation of oil droplets should be short, favouring higher speed ranges. For this reason, experiments were initially conducted at different load points and speeds. It was found that on the PC engine the operating point of 3000 rpm and an BMEP of 16 bar is well-suited for oil dosing experiments. This is because the compression end temperature is in the range of 500–\(550\,^{\circ } \hbox {C}\), aligning with the autoignition temperature window of oil droplets. Similarly, various operating points were investigated in advance on the HD engine to find an ideal operating point for oil dosing experiments. Through these investigations, the operating point of 1700 rpm and an BMEP of 12 bar were defined for the HD engine. This is because, in this case as well, the compression end temperature is in the same range like the PC engine and it runs almost free of pre-ignition. To make a qualitative assessment of the number of combustion anomalies before and after oil dosing, a test run with three phases of 20 min each was defined. The first phase serves for preconditioning, the second 20 min involve oil dosing, and the third phase is for post-conditioning.

In addition to defining the test run a measurement methodology focusing on the oil dosing tests was established and an evaluation tool was developed (Fig. 11).

The data are recorded throughout the entire test run at a rate of 10 Hz using the test bench automation program to monitor both, the experiment and the test engine, to ensure the plausibility of the test runs. From these measurement data, a temporal profile of oil consumption is also recorded and using the in Chapter 2 described carbon balance, for calculating the dosed oil quantity .

The indicating system continuously records fast data throughout the entire test run. This provides cyclic data for determining combustion anomalies from the entire cycles of a measurement sequence. In the subsequent analyses, the maximum cylinder pressure and the mass fraction burned 5 % (MFB05%) were used for each cycle, thus filtering out deviations caused by any anomalies. Therefore, it is possible in retrospect to present the exact number of pre-ignitions in each measurement phase and to validate the tests.

In addition to the propensity for oil-induced pre-ignitions, the developed analysis tool allows differentation between pre-ignitions that exhibit higher and lower levels of damage potential. The damage potential is highly dependent on the timing of the ignition onset (peak pressure and temperature). To account for this factor, a qualitative classification into “stronger” and “more moderate” pre-ignitions is made based on the MFB05%, with the spark timing serving as the boundary for the classification. In Fig. 12, the damage potential of different combustion anomalies, the crank angle-resolved cylinder pressure traces for a regular combustion (mean) and three oil-induced pre-ignitions (No. 18, 64, 105) are plotted, along with the spark timing. For the “stronger” pre-ignition No. 105, the MFB05% occurs significantly before the spark timing, resulting in an increase in peak pressure by up to 45 bar compared to regular combustion. In contrast, the ignition onset for the two “more moderate” pre-ignitions (No. 18 and 64) is noticeably later, leading to a lower level of damage potential to the crank mechanism. Furthermore, these very late pre-ignitions, like No. 18, were exclusively detected under the influence of engine oil. Droplet ignition is often cited in the literature as a possible triggering mechanism for oil-induced pre-ignitions [21, 22]. This would explain the delayed ignition due to the absence of self-ignition conditions for the oil at an earlier point in time. The compression temperature plays a major role here. Pre-ignitions triggered by hot spots or hot particles are typically induced much earlier [23]. However, it can also occur that, due to the consequential effect of an initial oil-induced pre-ignition, such as the significant heating of the combustion chamber or flanking deposits, a subsequent series of secondary pre-ignitions can occur, complicating the separated evaluation of the actual triggering mechanisms and the statistical comparison of different test oils.