Comparison of Turbulence Models with use of UnTRIM in the Delaware Bay

In this paper, various turbulence closure models are compared for use with a hydrodynamic 3D code in the Delaware Bay. More specifically, six (6) different turbulence closures, i.e. a constant eddy viscosity, an algebraic model, and 4 two-equation closure models have been used for comparison. These models have been implemented in the UnTRIM hydrodynamic code using the Generic Length Scale approach that mimics through its parameter combinations 3 of the two-equation closures (k-epsilon, k-omega, and k-kl) plus a separate Yamada Mellor 2.5 code. The UnTRIM code was used to model flows and salinity transport for a 2 month period by keeping the boundary and initial conditions unchanged for all model comparisons. The performance of each closure scheme was tested against salinity time series at a single station in the estuary. The comparisons reveal that the zero and first order approaches perform fair, but less accurate than the two-equation models that warrants the increased computational cost associated with these higher order approaches. Among the four 2-equation models, the k-epsilon model best represented the measured salinity time histories in the bay. The differences when compared to the other 2-equation models were significant enough to conclude that an arbitrary choice from the list of available or commonly used turbulence closure models is not sufficient. As a result, it is recommended that a test series be conducted to identify the best choice of turbulence closure in a specific application.