Revisiting the Gaussian Plume Modeling Paradigm: An Examination using High-Fidelity CFD Simulations over Complex Terrain

Despite having their origin in the nearly century-old work by O.G. Sutton, Gaussian plume models (GPMs) remain the industry standard for understanding and modeling atmospheric dispersion in regulatory and monitoring applications. The continued and widespread use of GPMs is enabled by their simple algebraic framework that relies on only a few meteorological parameters as inputs. Our interest in GPMs is motivated by the larger goal to mitigate climate change by accurately measuring and quantifying greenhouse gas (GHG) emissions in a quasi-continuous manner from the conventional energy sector. While GPMs do not account for the variation in landscape and obstacles near the emission source in their most basic form, both empirically determined and more physically justified modifications have been proposed over the years to account for their shortcomings. We perform eddy-resolving simulations that mimic the local unsteady surface and atmospheric conditions with the objective of introducing site-specific modifications to the GPM for improved prediction of concentrations. An immediate interest is near-surface dispersion of contaminants in oil & gas facilities where surface structures have heights comparable to or larger than corresponding sources. These simulations are intended to create a virtual yet physically accurate database that supplements simultaneous in-situ tracer gas measurements. A primary target is to develop a formal role for such computations in conjunction with instrumental measurements to better quantify total site emissions.

A set of detached eddy simulations (DES) with a resolution of 0.5 meters are performed in the first part of this investigation. The present computations construct the atmospheric boundary layer (ABL) under several representative conditions in and around the Methane Emissions Technology Evaluation Center (METEC) of Colorado State University. These representative conditions cover different atmospheric stability classes and multiple wind directions. The concept of eddy-viscosity in turbulence modeling simulations can be viewed as a product of characteristic velocity and length scales. Within this DES model, the length scale is clipped with the RANS length scale being provided by the k-ω model while the subgrid scales are modeled via a Smagorinsky-type model. The effect of surface roughness is included using a distributed forcing term in the momentum equation. Discrete point sources of passive scalar numerically approximated as 3D Gaussian functions are employed to introduce tracer gas flux into the computations. In the second part of this study, the distributed pollutant dispersion from a landfill facility is modeled using DES with surface resolution of 10 meters. The comparatively coarse grid is justified by our desire to instead focus on fence-line pollutant concentrations that are located at about 1 kilometer from the source. Information on surface topography near the landfill site is acquired using the database by the U.S. Geological Survey and incorporated into our CFD simulations using an immersed boundary method (IBM). The downstream pollutant concentrations are temporally tracked and statistically sampled to provide the time-averaged relative dispersion and meandering of the plume. Preliminary site-specific parameterization efforts are performed using a control-volume based approach and an established method that utilizes the two-point correlation function. They identify strong dependence of the plume dispersion parameters near the surface on local obstacles, especially for the METEC facility. As for the landfill site and barring the rapid changes in wind direction, initial results reflect the significance of capturing accurate large-scale surface variation in the downwind direction. A further in-depth analysis and comparison will be provided from estimates of the baseline GPM as well as Gaussian model formulations that account for near-surface effects.

Aside from offering insights on the accuracy and limitations of various GPMs in scenarios governed by the dispersion of industrial pollutants, the current scale-resolving simulations offer an independent and highly resolved spatio-temporal reference database that can be used to validate gas measurement sensors. It further provides promising training data that is facilitating the development of a generalized optimization approach for source characterization (flux quantification and source identification) of fugitive emissions.