The robustness of a large scale simulation is impacted by a series of factors from the quality of the mesh on which is performed to the accuracy of the numerical method and efficiency of the implementation. For non-linear problems such as Navier-Stokes simulations, which due to the nature of the equations are a stand-in for experiments, it is almost impossible to predict the outcome of a simulation. However a good mesh choice, a proven accuracy for a given solution set, and also efficient implementation choices render a given simulation superior to another. Present work is aimed at identifying and quantifying these issues in the code Nek5000 using the method of manufactured solutions, with applications to adjoint based optimization and mesh error analysis.