Uncertainty Analysis of Wideband Oxygen Sensor Based Strategy for Steady-State Biodiesel Blend Estimation

Real-time estimation and accommodation is critical for clean and efficient utilization of biodiesel blends in “fuel flexible” diesel engines. This paper utilizes a generalizable, physically-based, and experimentally-verified blend estimation strategy which uses exhaust oxygen sensor measurements coupled with engine control module (ECM) estimates of fuel and air flow to estimate the biodiesel blend fraction. The paper assesses the impact of uncertain variables on the estimation strategy. The strategy is essentially unaffected by biodiesel feedstock variations and, when applied to a Cummins 6.7-liter engine, is not susceptible to significant blend estimation discrepancies in response to expected fuel flow and oxygen sensor errors. However, observed errors in air flow estimates are expected to lead to large blend estimate errors. Use of direct air flow measurement or the use of a dynamic estimator (e.g., Kalman filter) synthesized from the model, the subject matter of future work, is expected to significantly reduce these errors.