During the development and optimisation of wave energy converters, numerical wave tanks are useful tools, providing detailed insight into the hydrodynamic performance of devices. Specifically, computational fluid dynamics (CFD)-based numerical wave tanks (CNWTs) can deliver high-fidelity, high-resolution results for a wide range of test conditions. However, CNWTs come at significant computational cost and require more man-hours during model setup, compared to lower-fidelity, frequency domain-based models. The computational costs can only be significantly decreased by improving the numerical solvers or by increasing expenditure on computational power. The required man-hours for the model setup, however, can be reduced by streamlining the setup of CNWTs. To this end, the formulation of best-practice guidelines can expedite this streamlining. A step toward such best-practice guidelines is blind tests. This paper presents the CNWT used for the authors’ contribution to the Collaborative Computational Project in Wave–Structure Interaction (CCP-WSI) Blind Test Series 3. In the employed numerical wave tanks, a self-calibrating impulse source wave maker is implemented for wave generation. In addition to the numerical results, and the comparison with the recently disclosed experimental data, the paper presents the spatial and temporal convergence studies, as well as results for the numerical wave maker calibration. The numerical results show average deviations with the experimental data of less than 10%. Furthermore, a correlation between the accuracy of the numerical replication of the wave and the agreement between numerical and experimental device motion is highlighted.