JETIVOS OPERATIVOS PARA EL AREA DE RECAUDACION

The effectiveness of the assistance for various quantities of injected air and com-pressed air supply is now investigated. The engine tests were repeated for three different set-points of the pressure regulator in the compressed air supply: 2.5, 3 and 3.5 bar. Similarly to the simulation-based assessment, the quantity of the injected air was not directly controlled. Instead, for various runs the weighting on the assist valve position in the MPC cost function was modified. Therefore, in a strict sense this procedure does not optimise the performance of the engine for a given quantity of air.

The mass of the injected air was calculated integrating the estimated flow through the injection valve. The flow, in turn, was calculated using the standard orifice equation, measured pressure ratio across the valve and the experimentally determined relation between the valve position and its effective area.

As the first measure of the response improvement, the acceleration time from 30 to 80 kph in 3^rd gear is considered. The results are shown in Figure 4.12.

As expected, a higher pressure in the supply allows for increased benefits in engine response. Injecting even a significant mass of air at the pressure of 2.5 bar reduced the acceleration time by only 0.2 seconds. It can be deduced that to achieve significant improvements, it is more effective to inject a given amount of air for a shorter time, but at a higher pressure.

0 0.01 0.02 0.03 0.04 0.05 7.4

7.6 7.8 8 8.2 8.4

Assist Air Mass (kg)

Time: 30−80 kph (s)

2.5 bar 3 bar 3.5 bar No Assist

Figure 4.12: Time from 30 to 80 kph as a function of injected quantity of air.

A relative reduction in time-to-torque is a common metric for assessing the transient response. It is defined as:

∆t%T,i= t_%T,i−t_%T,conv t%T,conv

(4.1) with t_%T,i being the time it takes from the initial level of torque up to a given percentage of the final torque for an investigated configuration and t_%T,conv being the similarly defined time for a reference system. In this case, the reference response is that of the engine without boost assistance. Three levels of torque were considered for this study: 50, 70 and 90% of maximum torque. The corresponding plots are shown in Figure 4.13.

The analysis of the presented figures leads to the conclusion that the reduc-tion in time to 50% of maximum torque is not significantly dependent on the pressure of air supply. The dependency is, however, visible in Figures 4.13b and 4.13c, where the improvements achieved with various supply pressures can be distinguished. These experimental results can be explained by the speed, up to which the turbocharger was accelerated in each case. The mass of air injected over a shorter period of time caused the turbocharger to spin up faster than the same amount injected over longer period of time. The turbocharger speed can be linked with the level of fuel injection via the smoke limit. Each of the considered supply pressures was sufficient to accelerate the turbocharger to quickly achieve

0 0.01 0.02 0.03 0.04 0

20 40 60 80 100

Assist Air Mass (kg) Reduction in Time to 50% Max Torque (%)

2.5 bar 3 bar 3.5 bar

(a) Time to 50% of maximum torque

0 0.01 0.02 0.03 0.04

0 20 40 60 80 100

Assist Air Mass (kg) Reduction in Time to 70% Max Torque (%)

(b) Time to 70% of maximum torque

0 0.01 0.02 0.03 0.04

0 20 40 60 80 100

Assist Air Mass (kg) Reduction in Time to 90% Max Torque (%)

(c) Time to 90% of maximum torque

Figure 4.13: Sensitivity of improvements to compressed air supply parameters.

50% of the maximum torque and therefore no distinction is observed in Figure 4.13a. The 2.5 bar supply cannot accelerate the turbocharger sufficiently quickly to achieve the 70% of the maximum torque as fast as the other supply levels.

Similarly, one needs 3.5 bar of supply to achieve 60% reduction in time to 90%

of maximum torque.

Bearing in mind the differences between the simulation model and the actual engine as well as certain differences in the details of test setups, the simulation results previously shown in Figure 4.5 appear to be consistent with experimental findings. Very comparable levels of transient response improvement were found using the parametric study conducted using the 1D model and during engine testing. The amount of air required to achieve these improvements was consid-erably lower during experimental verification: 60% improvement was achieved with about 0.06 kg of air in simulations and only 0.03 kg during experiments.

Further simulations revealed that it was a difference in the initial engine speeds that caused the difference in the amount of required air: tip-in with 1D model started at 850 RPM, whereas the initial engine speed used during engine tests was 1100 RPM, below which the Dual Mass Flywheel protection was limiting the response.

In this way, the simulation-based assessment of system capabilities is con-firmed to give realistic predictions of expected closed-loop system performance.