Research methodology

Experiment

To determine the downsizing potential of methanol, first, the maximum output torque of the engine with gasoline was tested through the test matrix following a design of experiment (DoE) approach. Two control parameters were selected, the valve timing and the intake boost pressure. To calculate the downsizing potential, the displacement of two engines which have the same engine power output is required [67]. In this study, tests were performed on a multi-cylinder engine, therefore the output (BMEP) is used to predict the downsizing potential. Based on the improvement of BMEP with methanol fuel, the downsizing possibility of the engine is calculated. The BMEP is further increased with higher boost pressure and different valve timing, therefore the maximizing potential can be estimated with a constraint of the maximum in-cylinder pressure. All of the tests were performed at an exhaust lambda of 1 for a high conversion efficiency of the TWC and minimum spark advance for best torque (MBT) or knock-limited spark advance (KLSA) ignition timing is used. The ignition timing was also limited by an exhaust gas temperature of 850^○C, to protect the turbine blades. The sweep tests of these two parameters then were carried out on methanol. For this sweep test, the ignition timing is fixed at the MBT ignition timing for the baseline case.

To achieve different valve overlap period, retarding exhaust valves or advancing intake valves or combined shifting of two valves can be used. For simplicity, simultaneous shifting of the two valves was performed in this research (which is called simultaneous variable valve timing, S-VVT). The intake valves are advanced, and the exhaust valves are retarded by the same valve shifting angle.

Because the original valve overlap is -30 CAD, the valve overlap with different valve shifting angle is calculated as:

valve overlap = 2 × valve shifting angle − 30 (2.1)

Figure 2.2 shows four examples of the valve strategies, baseline, S-VVT 10, S-VVT 25 and S-VVT 40. The number presents the valve shifting angle (advancement angle of the intake valves and retardation angle of the exhaust valves). The valve overlap in the case S-VVT 40 is 2×40 - 30 = 50 CAD.

90 180 270 360 450 540 630 Baseline

S-VVT 10 S-VVT 25 S-VVT 40

exhaustvalve intake valve

valve overlap

Crank angle (CAD aTDC_f) Figure 2.2: Valve events with different shifting strategies.

The influence of intake boost pressures was also studied. The actuating (gauge) pressure of the wastegate is 34 kPa. If the boost control system is deactivated, the maximum absolute pressure of the intake air is 1.34 bar. The MoTeC ECU can control to a higher value of the intake pressure; gauge pressures of 40 kPa, 60 kPa and 70 kPa were tested.

Numerical simulation

To determine the knock limit, the auto-ignition delay is needed. In this research, the auto-ignition delay of mixtures is predicted through a chemical kinetic simulation using Cantera [69]. The simulation was performed with a homogenous mixture in a constant volume reactor, and the auto-ignition delay is defined as the time when the maximum temperature rise rate is observed. The mechanism developed by Li et al. was used for methanol-air mixtures [70], and Mehl’s mechanism was used for gasoline-air mixtures [71]. A blend of iso-octane, n-heptane, and toluene with volumetric ratios of 35%, 15%, and 50% respectively was used to simulate gasoline. With that fraction, the toluene primary reference fuel (TPRF) has a similar RON and MON to the tested gasoline, 96.3 and 87.3 respectively [72].

The ignition delay ID (mass fraction burned 0-2%), combustion duration (10-75%

and 10-90% mass fraction burned, CA10-75 and CA10-90), unburned gas temperature, etc. were predicted through a three pressure analysis (TPA) in GT-Power [73]. The instantaneous intake, in-cylinder, and exhaust pressures were implemented into the model for the gas dynamic simulation. The mass flow rate of fuel was controlled to have the overall exhaust lambda of 1 and identical to the measured fuel flow rate (±2%). The TPRF in the prediction of ignition delay is

METHANOL AS A FUEL FOR A DIRECT-INJECTIONSIENGINE 29

also used to simulate gasoline.

Because the engine in this study is a direct injection engine, a cylinder evaporation model is required. Similar to the previous research on methanol [74], the mass fraction of the injected liquid fuel that vaporizes immediately after injection is 3%, and the duration for 50% fuel evaporated at 4000 rpm and liquid temperature of 600 K is 60 CAD. A shorter evaporation duration (20 CAD) is used for gasoline.

Because of methanol’s high latent heat, the evaporation is slower compared to gasoline surrogate fuel [75]. n-Heptane has a similar behavior in the mass fraction of vapour as gasoline at high temperature (T_gas = 353 K) [76], therefore the evaporation rate of n-heptane can be used to represent the evaporation of gasoline.

According to Lee and Law [77], the evaporation rate in a dry environment of n-heptane is 1.5 times higher than that of methanol. With the same mass, the evaporation duration for n-heptane (or gasoline) should be 40 CAD. Because the lower heating value of methanol is around half that of gasoline, the mass flow rate of gasoline is around 50% that of methanol. Therefore, the evaporation duration was taken to be 20 CAD. However, the vaporization heat taken from the walls was changed to 50% (from 75%). The reason is the injector location/direction. In the previous research, the fuel was injected from the side; thus it can be expected that more fuel impinges on the cylinder wall, and more heat is taken from the walls than in an engine which has direct injection from the centre of the pent-roof shaped cylinder head like the Volvo T3 engine.

The wall temperature is also an important parameter for the prediction of engine volumetric efficiency as well as the unburned gas temperature. One value was used for the cylinder head, cylinder liner, and the piston. Because no wall temperature measurement was available for this engine, a basic calculation for the wall temperature as a function of IMEP was performed, as follows [78]:

T_wall(K) =200

23IMEP(bar) +83 + 273.15 (2.2)

It is a primary estimation of the wall temperature, so makes no difference between fuels if the engine works at the same load. A constant temperature is used because a very small fluctuation of the wall temperature during the combustion (∼10 K) is typically found over the cycle [79]. The heat transfer model of Morel [47] was employed to calculate the heat losses to the wall. In order to match the cylinder pressure profile to the experiment, a heat transfer multiplier table for methanol was made for the baseline case. This heat transfer multiplier table then was used for other simulations of methanol. A different table for gasoline was also created.