One should guard against "the perfect being the enemy of the good". The ideal solution for the missing DDY is to get the long-term time series for the same location or station, and use that to calculate the values in the DDY, i.e., maximum dry-bulb and coincident wet-bulb temperatures, and vice-versa at 0.4%, 1.0%, 2.0%, and the minimum dry-bulb temperatures at 99.6% and 99% conditions. Unless one has access to such long-term data and software, it is unrealistic to expect that of the user or even weather data vender. If the weather file is synthetic data such as from MeteoNorm, there is no long-term time-series to speak of, anyway. Therefore, in such cases my suggestion is to calculate the DDY from the EPW, i.e., find the Cumulative Frequency of temperatures, which for hourly data is just a sorting of the temperatures from low to high, and then pick out the temperatures and coincident values at the various criteria (0.4%, 1%, 2% for cooling, and 99.6% and 99% for heating). If the EPW is a reasonably produced "typical year" file, trust me, the results would not be too far off (in the case of MeteoNorm, it's impossible to do any better, anyway) and probably better than using the DDY from another location elsewhere. If one is really worried about how "good" are such DDYs, you can try smoothing the CF curve. I don't want to dwell too much on the mechanics of deriving the design day values, which can get tedious, but I feel it's been unduly mystified.