The upcoming European legislation for heavy duty vehicles, EuVI, includes rules on monitoring of urea consumption and urea tank volume. These new rules put new demands on the level sensing system monitoring the urea tank.
The normal vehicle mounted liquid level sensing system today consists of a single, more or less vertically placed, sensor. When the vehicle is tilted the level of the fluid at the sensor position normally changes. This results in a measurement error in the calculated volume, as the volume is calculated based only on the information from the level sensor.
The presented modeling studies investigate the feasibility of using sensor fusion to improve the accuracy of liquid volume estimation on a heavy duty truck.
In the first study, the signals from multiple level sensors located in various liquid containing tanks on the truck are fused. Using realistic tank geometries, tank mounting repeatable and level sensor accuracies as input to a tolerance chain model enables the calculation of system responses to possible vehicle inclinations.
The model output is used to evaluate if it is possible to identify the vehicle tilt in two dimensions (front-rear and left-right) by the combined information collected from the change of fluid level.
When analyzing the change in fluid levels from the data it was found that the tilt can not be determined in a sufficiently reliable manner. The tilt combinations causing the fluid level to change in any, arbitrarily chosen, way could be more than one. This means that there are too many non unique permutations to determine the tilt and thus the fluid level with reasonable accuracy.
In the second study the sensor fusion between the inclination sensor of the brake system and the level sensors is investigated.
This approach is more promising but mounting tolerances of included components still result in an increased uncertainty and thus makes the use of the second concept challenging.