Computational simulations have become a widely used engineering tool. However, results of computational simulations are affected by modelling, input and numerical errors, which can rarely be determined due to the unavailability of the required true values. Therefore, input and numerical uncertainties need to be estimated to address the estimation of modeling errors . Numerical uncertainties have contributions of round-off, iterative and discretization errors and statistical errors for unsteady and/or stochastic simulations. For the statistically steady flows addressed in this study, the most significant contributions to the numerical uncertainty are originated by iterative and discretization errors, that have no guarantee of being independent contributions. On the other hand, estimation of input parameters uncertainties require uncertainty quantification techniques that may be extremely time consuming when sampling techniques are adopted. Therefore, it is not unusual to use coarse grids and loose iterative convergence criteria to estimate input parameters uncertainties. Naturally, this approach relies on the independency of input and numerical uncertainties. In this work, we estimate input uncertainties using sensitivity coefficients determined with finite-difference approximations, which is the simplest approach available. The test case is the flow around the KVLCC2 tanker at full scale Reynolds number calculated with the Reynolds-averaged Navier-Stokes equations using the k-w SST, an explicit algebraic Reynolds-stress and a Reynolds-stress model. Simulations are performed for significantly different levels of grid refinement and tolerances of the iterative convergence criteria and the uncertainty input parameters are the inlet turbulence intensity and the Reynolds number. Sensitivity coefficients are determined for the mean Cartesian velocity components and turbulence kinetic energy at the propeller plane. The results show that coarse grids and/or loose iterative convergence criteria can affect significantly the estimation of input parameters uncertainties.