Is this count data? If so a couple of ideas: You could turn it into pass/fail based upon your defect limit then convert your defect rate to a Ppk from a sigma-DPMO-Ppk table. You could also fit it as Poisson and use the distribution profiler to determine what your DPMO rate was (by setting the x axis to your spec limit and using the returned Y value) and convert that into a Ppk. The other option would be to turn your zero's into 0.0001's and then you'll have some additional distributions you can fit.