Cyclic variations and prior-cycle effects of ion current sensing in an HCCI engine: A time-series analysis. (15th April 2016)

• Record Type: Journal Article
• Authors: Chen, Yulin; Dong, Guangyu; Mack, J. Hunter; Butt, Ryan H.; Chen, Jyh-Yuan; Dibble, Robert W.
• Is Part Of: Applied energy. Volume 168(2016)
• Page Start: 628
• Page End: 635
• Publication Date: 2016-04-15
• Subjects: HCCI -- Cyclic variations -- Ion current sensing -- Fuel additives -- Symbol-sequence statistics; Power (Mechanics) -- Periodicals; Energy conservation -- Periodicals; Energy conversion -- Periodicals; 621.042
• DOI: 10.1016/j.apenergy.2016.01.126
• Languages: English
• ISSNs: 0306-2619

Abstract

Highlights: Nonlinear characteristics are identified to cause strong cyclic variations in ion current signals. Time series, return maps and CoV are applied to analyze cyclic variations of ion current signals. Due to the low ionization energy, the stability of ion signals can be largely improved by adding CsOAc. Pattern structures in prior cycles are determined by a symbol-sequence statistics method. By stronger deterministic features, ion current signals are more reliable to be predicted than pressure signals. Abstract: As an approach to replace pressure transducers, ion current sensing is a promising candidate for overcoming the difficult task of controlling the start of combustion in Homogeneous Charge Compression Ignition (HCCI) engines which require feedback from previous cycles. In this study, cyclic variations and prior-cycle effects of ion current signals are analyzed by comparing against pressure transducer signals using time-series methods in an HCCI engine. Additionally, the effects of various calibrated ion signal intensities are tested by adding cesium acetate (CsOAc) to the base fuel. Nonlinear characteristics of ion current signals are identified to cause strong cyclic variations through a single-zone model analysis with different equivalence ratios. By analyzing the time series, return maps, and coefficient of variations (CoV), the study finds that the stability of the ion signals can be largely improved by adding CsOAc due to the low ionization energy. After reconstructing a complex, nonlinear dynamical system model with symbol-sequence statistics, the measured cycle-resolved data of the ion current signal is analyzed to determine the pattern structures within prior cycles of fixed length, which is optimized by a modified Shannon entropy calculation. The results suggest that long, consecutive symbols of the ion current signal can be reliably predicted through the application of designed deterministic patterns especially when a small amount of CsOAc is added, although the ion current signal is normally considered a localized information provider and affected by many dynamical factors. Consequently, ion current signals are very promising for model-based control systems in HCCI engines with tolerable amounts of signal enhancing additives.