UC Berkeley

Air pollution detrimentally impacts environmental and human health. Specifically, communities living or working near seaports, intermodal rail yards, and other industrial activity, are exposed to greater levels of pollution and typically are made up of racial and ethnic minorities and populations of lower socioeconomic status. This relationship was fostered by historical, discriminating policies that have greatened exposure disparities and led to environmental injustices in the United States.In California (CA), pollution from diesel engines continues to disproportionately impact vulnerable populations. Diesel exhaust contains nitrogen oxides (NOx) and black carbon (BC), a major portion of diesel particulate matter (PM), which contribute to anthropogenic climate change and poor air quality. To reduce diesel pollution and its negative impacts on communities, the state government has endorsed two approaches. First, programs have been established to accelerate the adoption of diesel particle filters (DPFs) and selective catalytic reduction (SCR) for both on-road and off-road engines. Second, CA has passed legislation supplying funds for increased and assessable air quality monitoring in the most impacted communities as part of an effort to develop effective emission reduction plans. These approaches can benefit from the use of emerging low-cost technologies to facilitate broader monitoring opportunities and lead to greater impact across communities. As an example, lower cost sensors, if used in place of traditional research-grade analyzers, could enable wider application of the exhaust plume capture method, specifically as a monitoring tool to identify high-emitting trucks that may warrant inspection and maintenance. The exhaust plume capture method is an approach that can used to measure pollutants emitted by in-use diesel engines. The method can efficiently quantify emissions from a large sample of engines because of its non-invasive approach, which does not require installation of on-board measurement technologies. Emission rates (or factors) of several pollutants are quantify by measuring the chemical composition of an engine’s exhaust at a sampling site. The engine’s emissions profile can be linked to the specific attributes, such as engine model year or operating mode. This ultimately enables a detailed emissions characterization of an in-use on or off-road fleet, including the efficacy of emission control technologies and policies. However, low-cost sensors have for the most part only been evaluated under ambient conditions as opposed to source-influenced environments with rapidly changing pollutant concentrations. This dissertation compares BC emission factors determined using different BC and carbon dioxide (CO2) sensors that range in cost from $200–$20,000. Controlled laboratory experiments show that traditional zero and span steady-state calibration checks are not robust indicators of sensor performance when sampling short duration concentration peaks. Fleet BC emission factor distributions measured at two locations at the Port of Oakland in California with 16 BC/CO2 sensor pairs were similar, but unique sensor pairs identified different high-emitting trucks. At one location, the low-cost PP Systems SBA-5 CO2 sensor agreed on the classification of 90% of the high-emitters identified by the LI-COR LI-7000 CO2 monitor when both were paired with the Magee Scientific AE33 BC analyzer. Conversely, lower cost BC sensors when paired with the LI-7000 misclassified more than 50% of high emitters when compared to the AE33/LI-7000. Confidence in emission factor quantification and high-emitter identification improves with larger integrated peak areas of CO2 and especially BC. This work highlights that sensor evaluation should be conducted under application-specific conditions, whether that be for ambient air monitoring or source characterization. Further, as efforts to reduce off-road emissions continue, pollutant emissions rates from in-use vessels must be well characterized. This work explored the use of low-cost pollutant analyzers in a novel combination of exhaust plume capture with GPS location and speed data to quantified emission factors of BC and NOx from 21 engines on in-use excursion vessels and ferries operating in California’s San Francisco Bay. This produced ~60 fuel-based emission factors per engine including EPA uncertified and Tier 1–4 engines and across engine operating modes. BC and NOx emission factors (g kg-1) were lowest and least variable during fast cruising and highest during maneuvering and docked operation. SCR reduced NOx emissions by ~80% when functional. However, elevated NOx emissions that exceeded corresponding exhaust standards were measured on most Tier 3 and Tier 4 engines sampled, which can be attributed to inactive SCR during frequent low engine load operation. In contrast, BC emissions exceeded the PM emission standard for only one engine, and SCR systems employed as a NOx reduction technology also reduced emitted BC. Using these measured emission factors to compare commuting options, we show that the CO2-equivalent emissions per passenger-kilometer are comparable when commuting by car and ferry, but BC and NOx emissions can be several to more than ten times larger when commuting by ferry. Finally, low-cost air sensors offer the opportunity to deploy monitoring dense networks to characterize the spatial patterns in air pollution and identify local sources that most impacted communities. However, networks tend to measure only fine particulate matter (PM2.5) are unable to distinguish between primary and secondary PM2.5 pollution derived from either local or regional sources. To better understand local diesel pollution patterns, this work measured BC and PM2.5 during 4-week deployments in multiple seasons across 50 locations in Richmond, CA, a disadvantaged community in the Bay Area. This community is home to numerous sources of air pollution and heavy traffic from major roadways, such as the Chevron refinery, rail yards, and major highways. The network-average BC concentration in winter 2021 was 0.52 μg m-3, which was 3 times greater than the average of 0.17 μg m-3 measured in late spring. While winter BC concentrations ranged from 50% lower and higher than the network-average concentration, PM2.5 concentrations varied by only 20% of the network mean. A unique diurnal pattern of PM2.5 differed from that of BC in the winter and highlighted the influence of meteorology and local activity on primary and secondary pollution. Data from a collocated wind sensor with one BC sensor shows that the highest BC concentrations typically occur with the slowest winds in the winter. Further, the concurrent rise in nitrogen dioxide (NO2) and BC concentrations suggest that diesel engine emissions were also a contributing factor. In addition, sites ~100–500 meters from major roadways experienced higher levels of BC pollution in both the winter and spring, which included locations outside of three schools and one home. Notably, PM2.5 concentrations were less spatially variable than BC with residential, commercial, and industrial sites varying by only ±5% of the network-average concentration. During monitoring, wildfire smoke became a predominant pollutant and lead to 4- and 10-times higher BC and PM2.5 concentrations in the community, respectively. Block-by block differences in concentration and persistently polluted and clean locations were more evident for BC than for PM2.5 in this study. Thus, BC measurements may be more helpful than PM2.5 in terms of developing community emission reduction plans. Expanding the accessibility for communities to assess and evaluate their own air pollution data will lead to a greater understanding of personal exposure, enhanced community empowerment, and more informed policymaking to improve their ultimate health outcomes. Low-cost pollutant sensors are an excellent resource to provide this data and lead to more actionable steps in a community. Future use of low-cost sensors will aim to expand beyond air monitoring. There is a great potential to use data from low-cost sensors to validate emission inventories and models. Further, the portability and cost-effectiveness of this technology will allow for more personal monitoring to quantify pollutant exposure and derive dose-specific relationships with health outcomes. This work demonstrates the use of low-cost sensors to monitor and mitigate diesel emissions, which can only be achieved with proper technology evaluation, calibration, and deployment under the application-specific conditions. Comprehensive understanding of the abilities and limitations of these low-cost methods will ensure that effective and efficient strategies are developed to reduce emissions in the most vulnerable communities.