Transient Smoke Reduction Using a Hybrid Combination of Dimensional and Empirical Modeling 2013-01-0348
A hybrid calibration model combining dimensional and empirical modeling has been used to model transient smoke and demonstrate transient smoke reduction strategies during the turbocharger lag period of an electronically controlled heavy-duty diesel engine. This new hybrid approach termed the Non-Parametric Reduced Dimensionality approach (NPRD) uses GT-Power to transform the engine operating parameter model input space to a more fundamental, lower dimensional and less correlated model input space. A non-parametric nearest neighbor approach is then applied over this transformed model input space to make new predictions. The NPRD approach was used to predict transient FTP emissions of cumulative particulate matter (PM) within 7% of measured value, based solely on steady state training data. Conventional empirical methods extrapolated and produced unrealistic results over the same dataset because 40% of all transient points were classified as outliers as per the steady state training data. While comparing the NPRD model to measured data the transient EGR and fresh air flow rates were required for calculation of model inputs. Instead of relying on inaccurate ECM estimates of transient flow through the engine, GT-Power was used to calculate the flows rates based on measured boundary conditions. This utilized a new method of simulating transient data in which the measured manifold pressures and the EGR system flow resistance at every measured transient data point were replicated by quasi-static simulation by adjusting GT-Power actuators. The resulting actuator positions were then used for baseline transient simulations over which smoke reduction strategies were applied. Two simple approaches of controlling the EGR valve during the turbocharger lag period for smoke spike reduction with minimum NOx impact were justified and demonstrated. Both strategies are easy to incorporate within the engine control logic and are meant to be implemented in addition to existing smoke control strategies such as fuel-air- or Fuel-Oxygen-based smoke limits. Transient dimensional simulations in conjunction with the NPRD smoke predictor have been proposed as a tool for developing transient control and calibration strategies.