462496 Causal Analysis of the Environmental Distribution of Engineered Nanomaterials Using Bayesian Networks

Monday, November 14, 2016: 2:14 PM
Union Square 15 & 16 (Hilton San Francisco Union Square)
Muhammad Bilal1, H. Haven Liu2, Rong Liu1 and Yoram Cohen2, (1)Center for Environmental Implications of Nanotechnology, University of California, Los Angeles, Los Angeles, CA, (2)Chemical and Biomolecular Engineering, University of California, Los Angeles, Los Angeles, CA

The increasing use of engineered nanomaterials (ENM) in various products and applications is raising concerns regarding the possible releases of ENMs to the environment and potential environmental and health impact. It is therefore imperative to assess the expected ranges of ENM concentrations in various environmental media (e.g. air, water, soil, sediment) recognizing that ENMs can migrate across environmental phase boundaries. Thus, intermedia transport processes that are governed by various environmental factors are of significant importance when assessing potential ENM exposure levels. However, monitoring of ENMs exposure concentrations in multiple media is a daunting and costly endeavor. In this regard, modeling of ENMs releases and exposure concentrations can inform decision makers as to potential ENM emissions and concentration boundaries. In this regard, mechanistic models are most relevant to describing the distribution of ENMs in the multimedia environment. However, the use of such models for decision analysis is not trivial since multiple simulations may be required to be performed by experts in order to provide the information needed by decision analysis. In order to simplify the approach and make simulation data readily accessible for environmental decision analysis, a data-driven Bayesian Network (BN) model was developed for predicting environmental concentrations of ENMs (Al2O3, CeO2, Cu, SiO2, TiO2, and ZnO) over a wide range of environmental scenarios (in different geographical regions and for a range of ENM release scenarios) using data generated from a mechanistic multimedia model of the environmental distribution of ENMs. Using the BN model various environmental scenarios were rapidly evaluated and the impacts of input parameters on resulting ENM concentrations was assessed to provide clear causal relationships. For example, rain scavenging and dry deposition of airborne ENMs were the significant intermedia transport processes leading to the accumulation of ENMs in soil, water and sediment (where there was no direct ENM discharge to water). Through BN approach, it was demonstrated that rapid assessment of the dynamic mass distributions and exposure concentrations of ENMs can be conducted for a broad range of environmental scenarios of interest for direct input to environmental decision analysis.

Extended Abstract: File Not Uploaded
See more of this Session: Big Data and Sustainability
See more of this Group/Topical: Sustainable Engineering Forum