Model-based Development of Multi-Purpose Diagnostic Strategies for Gas Vehicles

Engines using compressed natural gas or liquefied petroleum gas are commonly equipped with control systems which are not yet able to completely monitor the gas supply line status. With a particular regard to safety but paying attention even to driving comfort and finally to polluting emissions reduction, two aspects in particular have been taken into account: the first one is the need to detect as soon as possible (and to react consequently) the presence of a problem occurring inside gas supply line (leakages and blocked-valves etcetera); the second one is the ability to detect an unsafe re-fuel operation, done with inserted ignition key, in order to switch off at least as more auxiliary loads as possible. The danger from such a manoeuvre may be identified in the high probability of an eventual electrostatic discharge and/or in the risk that the vehicle may be accidentally moved during the refilling operation.

Thanks to a model-based approach, these diagnosis may be carried out by recording and analyzing several signals already available on all common production systems and may allow to give some information even about entity and location of the leakages under different environmental and engine conditions. The whole strategy may be based either on “black” or on “white-grey” box algorithms, i.e. it may be performed through a system of experimental equations or by substituting some -or all- of them through physical models. By applying all the described procedures it would be also possible to commutate towards the other fuel (in case of dual-fuel engine equipped vehicles). Several experimental tests have been carried out (some more are planned for the immediate future), first in order to identify model parameters and finally to validate and demonstrate the effectiveness of proposed strategies. Some vehicles going to be produced in the immediate future will be already equipped with some of the described algorithms.