Delorme, A., Robert, J., Hollowell, W., Strobel, A. et al., "Fuel Economy Potential of Advanced AMT eCoast Feature in Long-Haul Applications," SAE Technical Paper 2014-01-2324, 2014, doi:10.4271/2014-01-2324.
In the recent years, Automated Manual Transmissions have become more popular for class 8 heavy trucks. Besides the benefits of smoother gear changes and reduced driver fatigue, AMTs can also greatly reduce fuel consumption by using optimized shifting strategies and advanced controls. The Detroit DT12 AMT demonstrated its ability to save fuel over a standard AMT, due in part to its eCoast feature. eCoast relies on intelligent and advanced electronic controls to safely allow the vehicle to coast on downgrades. While the engine is idling, the drag parasitic energy losses are decreased and the vehicle can fully use its momentum to travel further up and down hill. As one could expect, the type of route profile can greatly affect the fuel savings due to eCoast, since more hilly terrains might offer more opportunities to activate eCoast than flatter roads. In addition, when combined with different vehicle and driving parameters such as vehicle weight and driver desired cruise set speed, the fuel consumption reduction of eCoast is always there, but becomes a more complicated function. By using a detailed and validated vehicle model including the Detroit DT12 shifting and eCoast strategies implemented in Matlab/Simulink, this paper presents a comprehensive fuel consumption simulation study performed in Autonomie, a plug & play Matlab-based software environment and framework for automotive system simulation. Several US interstate routes, ranging from mostly flat to steep rolling hills, were selected and the corresponding high accuracy grade profiles were generated and used in these simulations. By considering a range of cruise set speeds and vehicle weights, this paper illustrates and identifies the optimal parameter combinations for each road profile and explains why the eCoast fuel savings relation to the simulation input variables can follow various trends.