Cpk and Ppk are designed to determine if a normal distribution process output, with a target value in the middle of the tolerance is
centered. Otherwise Cp or Pp identifies the probability of the normal distribution process output will make parts in spec. A unilateral tolerance cannot have a centered distribution, so Cpk and Ppk are not applicable. What this Cpk calculation gives you is a
distance from the specification - which is not the typical use of the index.
If you want to know the capability of a process with a physical limitation as a lower limit (such as 0 for runout) and a max tolerance above, you need to run a curve fit (you might assume gamma distribution if you are not familiar with curve fitting. The easiest way is to use Distribution Analyzer at variation.com) and determine it probability of being in tolerance based on the correct distribution model.