The paper further explores the application of computational fluid dynamics (CFD) codes for the study of the heat-transfer phenomena involved when working with fluids at supercritical pressure; bundle analysis is considered here in particular. As for previous simulations performed by the authors considering heat-transfer deterioration inside heated tubes, this application points out the limited capabilities of the most commonly used Reynolds-averaged Navier–Stokes models when approaching the heat-transfer deterioration phenomenon. It must be noted that some of the considered experimental conditions, which are very close to the pseudocritical temperature, represent at the same time one of the most challenging situations for the CFD codes and a very common situation if supercritical water-cooled reactors (SCWRs) will be developed. Improvements of the currently available turbulence models are then needed. The paper analyzes the most likely causes of the observed insufficient quality of the obtained predictions. In addition to comparing the measured and calculated wall temperature trends, the effect of the presence of the spacer grids on the turbulent flow is considered. Spacers are in fact very important to assure the structural stability of fuel, though they also affect the flow, generally improving the turbulence conditions in their neighborhood and slightly impairing it in the downstream region. A comparison between predictions performed including or not including the spacers is also performed.