Accuracy of boundary layer treatments at different Reynolds scales

Resistive forces associated to boundary layers (‘friction’) are usually out of scale in physical models of hydraulic structures, especially in the case of hydraulically smooth walls, generating distortions in the model results known as scale effects, that can be problematic in some relevant engineering problems. These scale effects can be quantified and corrected using suitable numerical models. In this paper the accuracy of using numerical simulation through the Reynolds Averaged Navier-Stokes (RANS) approximation in order to represent the head losses introduced by friction in hydraulically smooth walls is evaluated for a wide range of Reynolds scales. This is performed by comparing the numerical results for fully developed flow on circular pipes and between parallel plates against experimental results, using the most popular wall treatments. The associated numerical errors, mesh requirements and ranges of application are established for each treatment. It is shown that, when properly applied, RANS models are able to simulate the head losses produced by smooth wall friction accurately enough as to quantify the scale effects present in physical models. A methodology for upscaling physical model results to prototype scale, free of scale effects, is proposed