NON-INTRUSIVE SENSING SITE SELCECTION AND START OF COMBUSTION IDENTIFICATION IN MULTI-CYLINDER COMPRESSION-IGNITION ENGINES

Abstract

This study investigates the impact of sensor location on accelerometer-based sensing of combustion phasing for compression-ignition engines. Ten accelerometer locations were studied on a light-duty compression-ignition engine for a set of conditions with variations in engine load, speed, injection timing, and injection strategy. Start of combustion (SOC) was identified from the filtered acceleration signal using a previously developed approach. Each location was assessed using both signal-based metrics, including magnitude squared coherence (MSC) between block surface acceleration and in-cylinder pressure, as well as SOC outcome-based metrics, such as detection success rate. Results demonstrate that the mounting location has a significant impact on the ability to extract combustion phasing information from the accelerometer signal. Sensors mounted on the front face of the engine produced the strongest signals for an individual cylinder. For multi-cylinder sensing, side-mounted locations delivered the most reliable performance, with SOC detection success above 98 percent, defined as correctly identifying the acceleration peak most closely aligned with the corresponding pressure-derived SOC for each cycle. This work outlines a practical framework for selecting and evaluating accelerometer mounting locations, enabling broader use of accelerometers in engine platforms operating on a range of combustion approaches. Building on the results of the location study, additional efforts examine non-intrusive combustion sensing performance and interpretation under operating conditions beyond steady-state, traditional mixing-controlled diesel combustion. Sensing during transient operation, alternative sensing hardware (including a low-cost knock sensor), and sensing during energy-

assisted compression ignition (EACI) operation are investigated. Together, these results indicate that non-intrusive SOCa sensing can be extended to production relevant hardware and non-steady operation, but that the hardest cases are those with rapidly evolving or multi-stage heat release. Low-cost sensing hardware can remain effective at strong locations, transient load changes drive most miss identifications, and assisted combustion changes both the signal strength and the meaning of the detected SOCa.