Victor, it probably would be more accurate to increase your error on the conditions based on simpler assumptions of how the weather might change over the time rather than trying to model your building with predicted weather based on climate models.
The climate models will bring an error greater than the type you will find on an energyplus model, it can be hard to evaluate the influence of these errors on your model. I suggest you get a typical year epw and create some cenarios based on it (for example increase the yearly average maximum outside temperatures or decrease the overall moist in the air by 20%), and these cenarios you may define on climate prediction data and studies, in a way you work with probable cenarios and keep yourself on real world options.
If the typical year does not comply with what you need, another option is to get last year (or the last five years, I´ve done a similar thing for rain analysis on LEED building water use) observed weather data, create an epw based on this data and then create the cenarios from your epw.
Your objective is to evaluate building systems, systems sizing, comfort issues? Depending on the information you wish to find there can be different aproaches to it.
good luck,
Leandro Biondo
Environment Analyst / Meteorologist
Brazilian Forest Services - Environment Ministry