Estimation of the Stoichiometric Air-Fuel Ratio in Liquefied Petroleum Gas-Injected Engines

Air-fuel ratio control in gasoline engines has so far relied upon the fact that the stoichiometric air-fuel ratio of gasoline is an identified constant, largely thanks to its consistent chemical composition. In the case of Liquefied Petroleum Gas (LPG), chemical composition is subject to high variability due to geological and economic factors, amongst others. The implication of this variability is unpredictable stoichiometric air-fuel ratio of the fuel supply within any given vehicle, and ultimately degraded control of air-fuel ratio. This paper addresses the problem of stoichiometric air-fuel ratio estimation by evaluating the measurement and modeling of the relative permittivity of fuel, and also the method of iterative computation. For the estimation method proposed in this paper, simulation results are presented to demonstrate its effectiveness. INTRODUCTION For the automotive industry, the impetus of finding alternative fuels to gasoline has been driven by the everheightening environmental consciousness of society at large. Liquefied Petroleum Gas (LPG) has been identified as one of the most attractive alternative automotive fuels for its favourable environmental as well performance characteristics [1], [10]. However, air-fuel ratio control, an integral part of the exhaust minimization process, proves to be much more difficult in LPG engines than its gasoline counterpart. One of the main contributors to this problem is the fact that LPG composition is subject to high variability due to factors such as the geological characteristics of the crude oil extraction site, local market prices [1] and the mixing of different blends of LPG by motorists through refuelling. The aim of this paper is to help take advantage of LPG’s clean-burning nature by addressing the aforementioned problem. It must be mentioned that there exist thermodynamic issues which compound the air-fuel ratio control problem, i.e. the possibility of mixed-phase fuel being delivered to the injectors due to the vaporisation of lighter fuel components within the fuel line. However, though pertinent to the problem at large, it is assumed for the purposes of this paper that these characteristics are negligible and/or can be prevented through appropriate design modifications. The implication of unknown LPG composition is that the stoichiometric air-fuel ratio (the ratio at which postcatalyst pollutants are minimized and denoted here as S r ) is rendered unknown, contrary to gasoline for which it is a known constant. Thus, to minimize emissions by air-fuel ratio control, S r is used as the control input, where injected fuel mass ( f m ) is the control variable that is designed to achieve rS given an estimate of the mass airflow ( a m̂ ) into the cylinder. A survey of current air-fuel ratio control research [2], [4], [9], shows that current efforts largely do not encompass the case where the control reference is unknown. Though not stated, conventional closed-loop schemes can compensate the effects of an unknown reference, as the reference error is lumped together with the error in a m̂ . It is clear that this is a less-than-ideal solution as the air-fuel ratio controller is typically a fuel table which maps the control law to each quantized operating point of the engine (represented by the pair manifold air pressure m P (bar) and engine speed n (rev/min)), and therefore corrections at any one time cannot be applied over the entire table. On the other hand, by obtaining an estimate of S r , which is common throughout the table, a global update of the fuel table is then allowed. A further advantage of this is that air-fuel ratio controllers would not be unnecessarily burdened with the task of compensating for errors in S r by adjusting a m̂ during the drive cycle, giving rise to improved control performance, particularly in transient operation. In this paper, S r identification is derived from the initial conjecture that S r of any fuel blend may be inferred by measuring a physical property of the blend. The result of this investigation shows that for fuel blends exceeding two chemical components, the relationship between S r and the property of choice is underdetermined but bounded. The latter property of the mapping leads to the development of a computation scheme of S r estimation using the exhaust air-fuel ratio feedback. Downloaded from SAE International by University of Melbourne, Tuesday, January 07, 2014 06:16:50 PM