Membrane proteins are receiving an increase in attention due to their rolls in cell signaling and recognition. In particular, they are prime targets for drug design efforts. Computational prediction of binding affinity provides a useful prescreening and ranking tool for rational drug design, which may be attained using binding free energy calculations. Implicit solvent based methodologies provide efficient means of performing such computations. While generalized implicit membrane solvation has been mainly supported under Generalized Born based methodologies, which seek to approximate full Poisson-Boltzmann based computations, extension of numerical Poisson-Boltzmann solvers has occurred only relatively recently. Incorporation of implicit membrane models into the MM-PBSA framework in AMBER under the PBSA module required extension of existing accelerated linear solvers to allow support of periodic boundary conditions. Finally, the MM-PBSA implicit membrane methodology was demonstrated using the human purinergic platelet receptor (P2Y12R), for which structural and experimental binding data was recently released to the protein data bank. This included detailed examination of relevant parameters, such as choice of non-polar solvation term and selection of appropriate protein and membrane dielectric. Multi-trajectory methodology was also investigated briefly.