Development of a highly resolved 3-D computational model for applications in water quality and ecosystems

This dissertation presents the development and application of a computational model called BioChemFOAM developed using the computation fluid dynamic software OpenFOAM (Open source Field Operation And Manipulation). BioChemFOAM is a three dimensional incompressible unsteady-flow model that is coupled with a water-quality model via the Reynolds Average Navier-Stokes (RANS) equations. BioChemFOAM was developed to model nutrient dynamics in inland riverine aquatic ecosystems. BioChemFOAM solves the RANS equations for the hydrodynamics with an available library in OpenFOAM and implements a new library to include coupled systems of species transport equations with reactions. Simulation of the flow and multicomponent reactive transport are studied in detail for fundamental numerical experiments as well as for a real application in a backwater area of the Mississippi River. BioChemFOAM is a robust model that enables the flexible parameterization of processes for the nitrogen cycle. The processes studied include the following main components: algae, organic carbon, phosphorus, nitrogen, and dissolved oxygen. In particular, the research presented has three phases. The first phase involves the identification of the common processes that influence the nitrogen removal. The second phase covers the development and validation of the model that uses common parameterization to simulate the main features of an aquatic ecosystem. The main processes considered in the model and implemented in BioChemFOAM are: fully resolved hydraulic parameters (velocity and pressure), temperature variation, light's influence on the ecosystem, nutrients dynamics, algae growth and death, advection and diffusion of species, and isotropic turbulence (using a two-equation k-epsilon model). The final phase covers the application and analysis of the model and is divided in two sub stages: 1) a qualitative comparison of the main processes involved in the model (validation with the exact solution of different components of the model under different degrees of complexity) and 2) the quantification of main processes affecting nitrate removal in a backwater floodplain lake (Round Lake) in Pool 8 of the Mississippi River near La Crosse, WI.

The BioChemFOAM model was able to reproduce different levels of complexity in an aquatic ecosystem and expose several main features that may help understand nutrient dynamics. The validation process with fabricated numerical experiments, discussed in Chapter 4, not only presents a detailed evaluation of the equations and processes but also introduces a step-by-step method of validating the model, given a level of complexity and parameterization when modeling nutrient dynamics in aquatic ecosystems. The study cases maintain fixed coefficients and characteristic values of the concentration in order to compare the influences that increasing or decreasing complexity has on the model, BioChemFOAM. Chapter 4, which focuses on model validation with numerical experiments, demonstrates that, with characteristic concentration and coefficients, some processes do not greatly influence the nutrient dynamics for algae.

Chapters 5 and 6 discuss how BioChemFOAM was subsequently applied to an actual field case in the Mississippi River to show the model's ability to reproduce real world conditions when nitrate samples are available and other concentrations are used from typical monitored values. The model was able to reproduce the main processes affecting nutrient dynamics in the proposed scenarios and for previous studies in the literature. First, the model was adapted to simulate one species, nitrate, and its concentration was comparable to measured data. Second, the model was tested under different initial conditions. The model shows independence on initial conditions when reaching a steady mass flow rate for nitrate. Finally, a sensitivity analysis was performed using all eleven species in the model. The sensitivity takes as its basis the influence of processes on nitrate fate and transport and it defines eight scenarios. It was found in the present parameterization that green algae as modeled does not have a significant influence on improving nitrate spatial distributions and percentage of nitrate removal (PNR). On the other hand, reaction rates for denitrification at the bed and nitrification in the water shows an important influence on the nitrate spatial distribution and the PNR. One physical solution, from the broad range of scenarios defined in the sensitivity analysis, was selected as most closely reproducing the backwater natural system. The selection was based on published values of the percentage of nitrate removal (PNR), nitrate spatial concentrations, total nitrogen spatial concentrations and mass loading rate balances. The scenario identified as a physically valid solution has a reaction rate of nitrification and denitrification at the bed of 2.37x10-5 s-1. The PNR was found to be 39% when reaching a steady solution for the species transport. The denitrification at the bed process was about 6.7% of the input nitrate mass loading rate and the nitrification was about 7.7% of the input nitrate mass loading rate.

The present research and model development highlight the need for additional detailed field measurements to reduce the uncertainty of common processes included in advanced models (see Chapter 2 for a review of models and Chapter 3 for the proposed model). The application presented in Chapter 6 utilizes only spatial variations of nitrate and total nitrogen to validate the model, which limits the validation of the remaining species. Despite the fact that some species are not known a priori, numerical experiments serve as a guide that helps explain how the aquatic ecosystem responds under different initial and boundary conditions. In addition, the PNR curves presented in this research were useful when defining realistic removal rates in a backwater area. BioChemFOAM's ability to formulate scenarios under different driving forces makes the model invaluable in terms of understanding the potential connections between species concentration and flow variables. In general, the case study presents trends in spatial and temporal distributions of non-sampled species that were comparable to measured data.