Water-based condensation particle counters comparison near a major freeway with significant heavy-duty diesel traffic

Abstract: This study compares the instrumental performance of three TSI water-based condensation particle counter (WCPC) models measuring particle number concentrations in close proximity (15 m) to a major freeway that has a significant level of heavy-duty diesel traffic. The study focuses on examining instrument biases and performance differences by different WCPC models under realistic field operational conditions. Three TSI models (3781, 3783, and 3785) were operated for one month in triplicate (nine units in total) in parallel with two sets of Scanning Mobility Particle Sizer (SMPS) spectrometers for the concurrent measurement of particle size distributions. Inter-model bias under different wind directions were first evaluated using 1-min raw data. Although all three WCPC models agreed well in upwind conditions (lower particle number concentrations, in the range of 103–104 particles cm−3), the three models'' responses were significantly different under downwind conditions (higher particle number concentrations, above 104 particles cm−3). In an effort to increase inter-model linear correlations, we evaluated the results of using longer averaging time intervals. An averaging time of at least 15 min was found to achieve R 2 values of 0.96 or higher when comparing all three models. Similar results were observed for intra-model comparisons for each of the three models. This strong linear relationship helped identify instrument bias related to particle number concentrations and particle size distributions. The TSI 3783 produced the highest particle counts, followed by TSI 3785, which reported 11% lower during downwind conditions than TSI 3783. TSI 3781 recorded particle number concentrations that were 24% lower than those observed by TSI 3783 during downwind condition. We found that TSI 3781 underestimated particles with a count median diameter less than 45 nm. Although the particle size dependency of instrument performance was found the most significant in TSI 3781, both models 3783 and 3785 showed somewhat size dependency. In addition, within each tested WCPC model, one unit was found to count significantly different and be more sensitive to particle size than the other two. Finally, exponential regression analysis was used to numerically quantify instrumental inter-model bias. Correction equations are proposed to adjust the TSI 3781 and 3785 data to the most recent model TSI 3783. [Copyright &y& Elsevier]