Browsing by Author "Jathar, Shantanu, advisor"

Now showing 1 - 8 of 8

Open Access
Artificial neural networks for fuel consumption and emissions modeling in light duty vehicles
(Colorado State University. Libraries, 2019) Chenna, Shiva Tarun, author; Jathar, Shantanu, advisor; Bradley, Thomas, committee member; Anderson, Chuck, committee member
There is growing evidence that real world, on-road emissions from mobile sources exceed emissions determined during laboratory tests and that the air quality, climate, and human health impacts from mobile sources might be substantially different than initially thought. Hence, there is an immediate need to measure and model these exceedances if we are to better understand and mitigate the environmental impacts of mobile sources. In this work, we used a portable emissions monitoring system (PEMS) and artificial neural networks (ANNs) to measure and model on-road fuel consumption and tailpipe emissions from Tier-2 light-duty gasoline and diesel vehicle. Tests were performed on at least five separate days for each vehicle and each test included a cold start and operation over a hot phase. Routes were deliberately picked to mimic certain features (e.g., distance, time duration) of driving cycles used for emissions certification (e.g., FTP-75). Data were gathered for a total of 49 miles and 145 minutes for the gasoline vehicle and 52 miles and 165 minutes for the diesel vehicle. Fuel consumption and emissions data were calculated at 1 Hz using information gathered from the vehicle using the onboard diagnostics port and the PEMS measurements. Route-integrated tailpipe emissions did not exceed the Tier-2 emissions standard for CO, NOX, and non-methane organic gases (NMOG) for either vehicle but did exceed so for PM for the diesel vehicle. We trained ANN models on part of the data to predict fuel consumption and tailpipe emissions at 1 Hz for both vehicles and evaluated these models against the rest of the data. The ANN models performed best when the training iterations (or epochs) were set to larger than 25 and the number of neurons in the hidden layer was between 7 and 9, although we did not see any specific advantage in increasing the number of hidden layers beyond 1. The trained ANN model predicted the fuel consumption over test routes within 5.5% of the measured value for both gasoline and diesel vehicles. The ANN performance varied significantly with pollutant type for the two vehicles and we were able to develop satisfactory models only for unburned hydrocarbons (HC) and NOX for diesel vehicles. Over independent test routes, the trained ANN models predicted HC within 12.5% of the measured value for the gasoline vehicle and predicted NOX emissions within 3% of the measured values for the diesel vehicle. The ANN performed better than, and hence could be used in lieu of, multivariable regression models such as those used in mobile source emissions models (e.g., EMFAC). In an 'environmental-routing' case study performed over three origin-destination pairs, the ANNs were able to successfully pick routes that minimized fuel consumption. Our work demonstrates the use of artificial neural networks to model fuel consumption and tailpipe emissions from light-duty passenger vehicles, with applications ranging from environmental routing to emissions inventory modeling.
Embargo
Autonomous low-cost network of ozone sensors: to study the spatial distribution and exposure in underserved agricultural communities in central California
(Colorado State University. Libraries, 2025) Gunniah Vijayakumar, Akshay Kumar, author; Jathar, Shantanu, advisor; L'Orange, Christian, committee member; Magzamen, Sheryl, committee member; Quinn, Casey, committee member; Carter, Ellison, committee member
Ozone (O3), a criteria air pollutant, is often overlooked in rural and remote regions, leaving the spatial distribution and exposure levels poorly understood, particularly in underserved communities. In this study, we developed and deployed a network of 12 autonomous, low-cost, and solar-powered air quality monitoring units (VOZboxes) in California's San Joaquin Valley (SJV). Co-located with a reference monitor in Fresno, CA, the VOZboxes underwent calibration before and after field deployment in June and November 2023, respectively, to measure O3 concentrations over a dynamic range of 20 to 100 ppbv with an RMSE < 5 ppbv. Deployed across 11 unique locations in SJV from July to November 2023 at varying periods, the VOZboxes revealed modest spatial variability in O3 mixing ratios, with elevated concentrations recorded in bigger cities and smaller eastern townships, while lower concentrations were found in smaller western regions. By leveraging multivariate regression models for data calibration, the VOZboxes effectively assessed compliance with the national ambient air quality standard (NAAQS) for O3 (maximum daily 8-hour average of 70 ppbv) across locations. This study underscores the potential of low-cost environmental sensors for characterizing air quality and O3 exposure in rural and remote environments. Additionally, it emphasizes their utility as tools for addressing the monitoring needs of underserved communities and acts as a tool for environmental justice.
Open Access
Autonomous low-cost ozone sensors: development, calibration, and application to study exposure and spatial gradients
(Colorado State University. Libraries, 2022) Giardina, Dylan M., author; Jathar, Shantanu, advisor; Magzamen, Sheryl, committee member; Volckens, John, committee member; Bechara, Samuel, committee member
Ozone (O3), a criteria pollutant and atmospheric oxidant, is not routinely measured in rural and remote environments and hence exposure to ozone pollution in these regions remains poorly understood. In this work, we built, calibrated, and deployed five low-cost, autonomous ozone sensor systems (called MOOS) in Northern Colorado, a region that is non-compliant for O3 during the summertime. Each MOOS included the following components: (i) an Aeroqual SM50, a heated metal oxide ozone sensor, mounted inside a custom radiation shield, (ii) a power system that consisted of a 30 W solar panel, 108 Wh lithium-ion battery, and charge controller, (iii) a Particle Boron to acquire, process, and transmit data to the Cloud, and (iv) an environmental sensor to measure temperature, relative humidity, and pressure. In a three-week long collocated study, we found that all MOOS, calibrated using 48 hours of reference data, compared well against reference monitors with a measurement error between 4-6 parts per billion by volume (ppbv). Manufacturer- and laboratory-based calibrations over- and under-estimated ozone levels at higher and lower ozone mixing ratios, respectively. When deployed in Northern Colorado for an additional three weeks to measure O3 exposure and study O3 trends across an urban-rural gradient, we found that the MOOS, calibrated using data from the collocated study and calibrated using 48 hours of reference data in the field, demonstrated good sensor performance (RMSE of 3.98 - 8.80 ppbv and MBE of 0.22 - 3.82 ppbv). Compared to the collocated study, the field study resulted in larger measurement errors for all five MOOS (RMSE of 3.66 - 4.00 versus RMSE of 3.98 - 8.80). Furthermore, there was modest variability in the field performance across the different MOOS (RMSE < 5 ppbv) that could not be explained by environmental differences between the different sites (e.g., proximity of the MOOS to the reference monitor, land use type, temperature). We found that MOOS were able to capture 100% of non-compliant O3 days during the collocated study and between 25-87% of non-compliant O3 days during the field study depending on the calibration approach used. Furthermore, both reference monitors and MOOS deployed along the east-west corridor in Northern Colorado were able to capture the negative, west-east O3 gradients observed in previous aircraft and modeling studies. Overall, our study indicates that the MOOS shows promise as a low-cost O3 sensor that could be used to supplement routine ambient monitoring and characterize regional ozone pollution.
Open Access
Modeling the formation and composition of secondary organic aerosol from diesel exhaust using parameterized and semi-explicit chemistry and thermodynamic models
(Colorado State University. Libraries, 2017) Eluri, Sailaja, author; Jathar, Shantanu, advisor; Volckens, John, committee member; Pierce, Jeffrey, committee member; Farmer, Delphine, committee member
Laboratory-based studies have shown that diesel-powered sources emit volatile organic compounds that can be photo-oxidized in the atmosphere to form secondary organic aerosol (SOA); in some cases, this SOA can exceed direct emissions of particulate matter (PM); PM is a criteria pollutant that is known to have adverse effects on air quality, climate, and human health. However, there are open questions surrounding how these laboratory experiments can be extrapolated to the real atmosphere and how they will help identify the most important species in diesel exhaust that contribute to SOA formation. Jathar et al. (2017) recently performed experiments using an oxidation flow reactor (OFR) to measure the photochemical production of SOA from a diesel engine operated at two different engine loads (idle, load), two fuel types (diesel, biodiesel) and two aftertreatment configurations (with and without an oxidation catalyst and particle filter). In this work, we will use two different SOA models, namely the volatility basis set (VBS) model and the statistical oxidation model (SOM), to simulate the formation, evolution and composition of SOA from the experiments of Jathar et al. (2017). Leveraging recent laboratory-based parameterizations, both frameworks accounted for a semi-volatile and reactive POA, SOA production from semi-volatile, intermediate-volatility and volatile organic compounds (SVOC, IVOC and VOC), NOx-dependent multigenerational gas-phase chemistry, and kinetic gas/particle partitioning. Both frameworks demonstrated that for model predictions of SOA mass and elemental composition to agree with measurements across all engine load-fuel-aftertreatment combinations, it was necessary to (a) model the kinetically-limited gas/particle partitioning likely in OFRs and (b) account for SOA formation from IVOCs (IVOCs were found to account for more than four-fifths of the model-predicted SOA). Model predictions of the gas-phase organic compounds (resolved in carbon and oxygen space) from the SOM compared favorably to gas-phase measurements made using a Chemical Ionization Mass Spectrometer (CIMS) that, qualitatively, substantiated the semi-explicit chemistry captured by the SOM and the measurements made by the CIMS. Sensitivity simulations suggested that (a) IVOCs from diesel exhaust could be modeled using a single surrogate species with an SOA mass yield equivalent to a C15 or C17 linear alkane for use in large-scale models, (b) different diesel exhaust emissions profiles in the literature resulted in the same SOA production as long as IVOCs were included and (c) accounting for vapor wall loss parameterizations for the SOA precursors improved model performance. As OFRs are increasingly used to study SOA formation and evolution in laboratory and field environments, there is a need to develop models that can be used to interpret the OFR data. This work is one example of the model development and application relevant to the use of OFRs.
Open Access
Multi-day evolution of organic aerosol mass and composition from biomass burning emissions
(Colorado State University. Libraries, 2023) Dearden, Abraham C., author; Jathar, Shantanu, advisor; Bond, Tami, committee member; Pierce, Jeffrey, committee member
Biomass burning is an important source of primary and secondary organic aerosol (POA, SOA, and together, OA) to the atmosphere. The photochemical evolution of biomass burning OA, especially over long photochemical ages, is highly complex and there are large uncertainties in how this evolution is represented in models. Recently, we performed photooxidation experiments on biomass burning emissions using a small environmental chamber (~150 L) to study the OA evolution over multiple equivalent days of photochemical aging. In this work, we use a kinetic, process-level model (SOM-TOMAS; Statistical Oxidation Model-TwO Moment Aerosol Sectional) to simulate the photochemical evolution of OA in 18 chamber experiments performed on emissions from 10 different fuels. A base version of the model was able to simulate the time-dependent evolution of the OA mass concentration and its oxygen-to-carbon ratio (O:C) at short photochemical ages (0.5 to 1 equivalent days) but substantially underestimated the enhancement in both metrics at longer photochemical ages (>1 equivalent day). The OA after several days of equivalent photochemical aging was dominated by SOA (58%, on average) with the remainder being POA (42%, on average). Semi-volatile organic compounds, oxygenated aromatics, and heterocyclics accounted for the majority (86%, on average) of the SOA formed. Experimental artifacts (i.e., particle and vapor wall losses) were found to be much more important in influencing the OA evolution than other processes (i.e., dilution, heterogeneous chemistry, and oligomerization reactions). Adjustments to the kinetic model seemed to improve model performance only marginally indicating that the model was missing precursors, chemical pathways, or both, especially to explain the observed enhancement in OA mass and O:C over longer photochemical ages. While far from ideal, this work contributes to a process-level understanding of biomass burning OA that is relevant for its evolution at regional and global scales.
Open Access
Predicting hybrid vehicle fuel economy and emissions with neural network models trained with real world data
(Colorado State University. Libraries, 2017) Galang, Abril, author; Jathar, Shantanu, advisor; Bradley, Thomas, advisor; Anderson, Chuck, committee member
Physics-based hybrid vehicle simulation models for fuel economy (FE) exist but are computationally and financially expensive. These models simulate aspects of real-world drive cycles that include the driving environment, thermal management, driver input, and powertrain component behavior. In this study, an alternative method of hybrid vehicle FE simulation is developed by training and testing a time series neural network (NN) model using real world, on-road data. This enables NN models to model many aspects of on-road vehicle dynamics, like regular traffic stops, turning, and irregular accelerations and stops. Unlike the physics-based models, NNs have the advantage of lower computational costs, which could be utilized in near-real-time vehicle system control to determine optimal velocity planning and powertrain control. Models trained in this study used velocity-time traces as an input to predict instantaneous FE. The NN model predicted fuel economy within a mean absolute error of 0-5% for on-road measurements over a 40 minute, real world, city and highway drive cycle. NN models trained with varying lengths of datasets did not improve with training data longer than 35 minutes. When trained with this method, NN models were accurate when tested with data from multiple days of tests and various drive cycles. Multiple NN models were also trained with hybrid vehicles with varying control system settings. NNs can only successfully model a vehicle whose control system settings reflect the training of the model. These results are expected to improve with more comprehensive drive cycle data that includes data from different elevations and various climatic conditions. The predictions from the FE NN model were compared against predictions from the physics-based Autonomie model and a custom HEV simulation model developed at Colorado State University. NNs outperform these models when tested with on-road data to predict FE of a known vehicle. Using a portable emissions monitoring system (PEMS), NN models were also able to predict nitrous oxides and particulate matter emissions with <5% mean absolute error. The NN model method could be used to improve emissions estimates by capturing differences between real world and laboratory tested emissions. Recording and including more data from the vehicle and devices like the PEMS could further improve these NN models.
Open Access
Secondary organic aerosol formation from volatile chemical product emissions: parameters and contributions to anthropogenic aerosol
(Colorado State University. Libraries, 2023) Sasidharan, Sreejith, author; Jathar, Shantanu, advisor; Volckens, John, committee member; Pierce, Jeffrey, committee member
Volatile chemical products (VCP) are an increasingly important source of hydrocarbon and oxygenated volatile organic compound (OVOC) emissions to the atmosphere, and these emissions are likely to play an important role as anthropogenic precursors for secondary organic aerosol (SOA). While the SOA from VCP hydrocarbons is often accounted for in ambient air quality models, the formation, evolution, and properties of SOA from VCP OVOCs remains uncertain. We use environmental chamber data and a kinetic model to develop SOA parameters for ten OVOCs representing glycols, glycol ethers, esters, oxygenated aromatics, and amines. Model simulations suggest that the SOA mass yields for these OVOCs are on the same magnitude as widely studied SOA precursors (e.g., long-chain alkanes, monoterpenes, and single-ring aromatics) and these yields exhibit a linear correlation with the difference between the carbon and oxygen numbers of the precursor. When combined with emissions inventories for two megacities in the United States (US) and a US-wide inventory, we find that VCPs form 0.8-2.5× as much SOA, by mass, as mobile sources. Hydrocarbons (terpenes, branched and cyclic alkanes) and OVOCs (terpenoids, glycols, glycol ethers) make up 60-75% and 25-40% of the SOA arising from VCP use, respectively. This work contributes to the growing body of knowledge focused on studying VCP VOC contributions to urban air pollution.
Open Access
Simulating cut to length forest treatment effects on fire behavior over steep slopes
(Colorado State University. Libraries, 2023) Pittman, Kyle Tait, author; Jathar, Shantanu, advisor; Hoffman, Chad, advisor; Linn, Rod, committee member; Windom, Bret, committee member; Wei, Yu, committee member
The increase of wildfire size and behavior in many western U.S. forests is due to increased fuel loads resulting from the past century's fire suppression, logging, and grazing policies of the 20th century, along with compounding climactic changes including increased drought and temperatures. Fuel hazard treatments are the key land management tool used to reduce fire intensity and severity however these treatments are often not possible on steep terrain of over 30% slope. Cable tethered cut to length machinery opens new avenues for managers to treat forests in steep slopes, but there is limited data on how effective the treatments will be. I conducted a numerical experiment using the wildfire model, FIRETEC, coupled with the atmospheric dynamics model, HIGRAD, to understand the complex interactions of wind, topography, and fire behaviors of two cut to length forest treatments on slopes of up to 60%. Results show that treatments can effectively reduce some fire behaviors such as heat release and canopy consumption when compared to untreated forests on slopes. However, increased sub canopy wind penetration along the slopes following treatments results in marginal fire severity reduction regarding biomass consumption and variable results on rates of spread. The results of these numerical experiments indicate that CTL treatment can effectively reduce some fire behavior and severity, however the effects were marginal and additional research is needed to better understand treatment's effects.