Evaluation of an On-Board, Real-Time Electronic Particulate Matter Sensor Using Heavy-Duty On-Highway Diesel Engine Platform

California Air Resources Board (CARB) has instituted requirements for on-board diagnostics (OBD) that makes a spark-plug sized exhaust particulate matter (PM) sensor a critical component of the OBD system to detect diesel particulate filter (DPF) failure. Currently, non-real-time resistive-type sensors are used by engine OEMs onboard vehicles. Future OBD regulations are likely to lower PM OBD thresholds requiring higher sensitivity sensors with better data yield for OBD decision making. The focus of this work was on the experimental evaluation of a real-time PM sensor manufactured by EmiSense Technologies, LLC that may offer such benefits. A 2011 model year on-highway heavy-duty diesel engine fitted with a diesel oxidation catalyst (DOC) and a catalyzed DPF followed by urea-based selective catalytic reducer (SCR) and ammonia oxidation (AMOX) catalysts was used for this program. With the aid of an exhaust bypass, sensors were examined for their accuracy and variability at emission level of ~ 0.02 g/hp-hr (mid-way between Laboratory and OBD thresholds) using multiple repeats of various transient drive cycles. Seven sensors were benchmarked simultaneously against state-of-the-art laboratory particle instrumentation capable of measuring real-time soot mass and solid particle number and size.

Sensor response was compared with laboratory reference instruments over different window lengths of the transient cycles. Using 100 second windows, correlation between sensors and reference soot mass concentration for all transient tests (unit being mg/m³) yielded R² ~ 0.72 and standard error of estimate (SEE) ~ 29% (% of mean value). Similar analysis between sensors and reference flux (unit being grams/second) yielded R² of 0.72 and SEE of 39%. When the entire cycle length was used as window length, correlation between sensors and reference concentration yielded R² ~ 0.98 with a SEE ~ 6.9% (% of mean value). Correlation between sensors and reference flux yielded R² ~ 0.95 with a SEE ~ 11.2% (% of mean value). Results suggest that these sensors have tremendous potential to serve as exhaust emission monitors for a wide range of transient PM emissions using the 100-second windowing approach for improved accuracy.