Ion acceleration in the MeV range can be routinely achieved with table-top laser technology. One of the current challenges is to improve the energy coupling from the laser to the proton beam without increasing the laser peak power. Introducing nanostructures at the front target surface was shown to be beneficial for an efficient transfer of energy to the electrons. In this manuscript, we study by using full-scale three-dimensional particle-in-cell simulations and finite laser pulses, the process when a proposed optimal target with triangular nanostructure (previously found to allow 97% laser energy absorption) is used . We demonstrate that the absorbed laser energy does not depend on the dimensionality in the range of parameters presented. We also present an analytical model for laser absorption that includes deviations from the ideal conditions. This is supported by a numerical parameter study that establishes the tolerance with respect to the nanostructure size, use of different ion species, existence of preplasma, etc. We found that altering the target thickness or using different ions does not affect the absorption, but it does affect the energy redistribution among the different plasma species. The optimal configuration ($h = 1~lambda,~ w = 0.7~ lambda$) is robust with respect to the target fabrication errors. However, high contrast laser pulses are required, because a pre-plasma layer with a thickness on the order of 0.5 lambda is enough to lower the laser absorption by more than a 10% in a non-optimal scenario.