Helical steam generators are proposed for use in a number of advanced nuclear reactor designs. The cross-flow around the helical tubes is a complex flow-field, and accurate knowledge of this flow is necessary for estimating pressure drop, heat transfer, and risk of flow-induced vibration. However, legacy data for helical tube cross-flow are scarce, and building new large-scale experiments that investigate relevant phenomena can be costly. Thus large uncertainties must currently be taken into account in the design of these systems. Numerical modeling with CFD can provide improved insight into the flow phenomena to reduce this uncertainty, but choosing a methodology can prove difficult. LES methods provide high-fidelity data, but require immense computational time to perform even an investigatory calculation for a moderate-sized sector, let alone for many design iterations. URANS methods offer significantly lower computational time, but it can be difficult to confidently justify the accuracy of a particular model without validation, particularly given the highly three-dimensional and complex flow-field present here. To better establish a basis for URANS turbulence modeling, an LES simulation was performed using Nek5000, a massively-parallel spectral element code developed at Argonne National Laboratory, for the geometry of a legacy helical tube bundle experiment. Data from this high-fidelity LES simulation were compared with URANS simulations using a number of turbulence models with the commercial code STAR-CCM+. Turbulent kinetic energy in the flow channels as well as bundle pressure drop were compared. The values of these key parameters were found to vary significantly between different turbulence models, with some models predicting pressure drops and kinetic energies well below those seen in LES. Some models were identified that showed good potential for predicting helical tube bundle flow phenomena. Further work, at a wider range of flow velocities, will be useful to further solidify the range of applicability of these models.