A Computational Fluid Dynamics Analysis of Spatial Roughness Distributions on Ship Resistance

36th Symposium on Naval Hydrodynamics (SNH36, 2026)
Sang-seok Han 1*, Momchil Terziev 1, Tahsin Tezdogan 2
1 Department of Naval Architecture, Ocean and Marine Engineering, University of Strathclyde, the United Kingdom
2 Department of Civil, Maritime and Environmental Engineering, University of Southampton, the United Kingdom

Abstract

In ship hydrodynamics, hull surface roughness modifies near-wall turbulence and boundary-layer development, thereby influencing resistance along the hull. In most resistance prediction methodologies, these effects are represented using an equivalent uniform roughness height. In practice, however, ship hulls exhibit non-uniform roughness distributions due to biofouling development and coating degradation. The hydrodynamic implications of this spatial variability are not yet fully understood.

This study presents a systematic computational fluid dynamics (CFD) investigation of spatially distributed hull roughness effects on ship resistance. Two benchmark hull forms—the KRISO Container Ship (KCS) and the KRISO Very Large Crude Carrier 2 (KVLCC2)—are analysed. Ten idealised roughness distributions are considered, including uniform, linear, non-linear, stochastic, and shear stress-based configurations. All cases are constructed to preserve a reference-equivalent roughness severity, such that only the influence of spatial roughness distribution is isolated.

The results show that, under identical reference-equivalent roughness severity, measurable differences in the total resistance coefficient CT arise when only the spatial distribution of hull roughness is varied. The resulting variations in CT between idealised roughness distributions are more pronounced for the fuller KVLCC2 hull form. Flow-field analysis indicates that these variations in CT are associated with cumulative boundary-layer development and wake recovery, with a strong sensitivity to roughness placement in the lower hull region. The findings suggest that the spatial distribution of hull roughness plays a key role in governing boundary-layer development and the resulting resistance response, even when the overall roughness severity is unchanged. Explicit consideration of roughness distribution therefore provides a more physically consistent basis for interpreting roughness-induced resistance effects.

Conference paper (2026)
36th Symposium on Naval Hydrodynamics (SNH36)
Proceedings forthcoming
Selected Figures
Distribution of y-plus minus R-plus over the wetted hull surface of the KVLCC2 for representative roughness cases
Figure 1. Distribution of y+ − R+ over the wetted hull surface of the KVLCC2 for representative roughness cases. From top to bottom: (a) uniform, (b) linear–steep, (c) non-linear–steep, (d) random, (e) direct-shear, and (f) inverse-shear stress-based cases.
This figure verifies the near-wall resolution used for the roughness modelling on the KVLCC2 hull. Representative uniform, gradient-based, random, and shear-stress-based roughness cases are shown to illustrate that the wall-function treatment remains consistent across the main roughness-distribution strategies considered in the study.
Decomposition of total resistance into frictional and pressure components for the KVLCC2 hull under the B20 percent reference-equivalent roughness severity
Figure 2. Decomposition of total resistance into frictional and pressure components for the KVLCC2 hull under the B20% reference-equivalent roughness severity.
The resistance decomposition shows that frictional resistance remains the dominant contribution to total resistance for the roughened KVLCC2 hull, while the pressure component still plays a non-negligible role. The comparison across roughness distributions helps explain why the fuller KVLCC2 hull exhibits stronger sensitivity to spatial roughness redistribution than the KCS hull.
Axial velocity field and relative axial velocity deviations around the KVLCC2 hull, extracted at y equals 0.006 Lpp
Figure 3. Axial velocity (Vx/Vship) field and relative axial velocity deviations around the KVLCC2 hull, extracted at y = 0.006Lpp.
Panel (a) shows the uniformly roughened reference case. Panels (b)–(d) show relative axial velocity deviations from the uniform reference for the linear–steep, random, and direct-shear stress-based distributions, respectively. The figure illustrates how different spatial roughness distributions modify wake development around the KVLCC2 hull, with the linear–steep case showing faster wake recovery, the random case remaining relatively close to the uniform reference, and the direct-shear case producing more localised deviations near the stern and wake region.