RPDs are typically calculated when both sample results are greater than 5 times the reporting detection limit (RDL).
When one or both of the sample results are <5x RDL the RPD is reported as "NC" (not calculable).
As the measured result for duplicate samples approaches the RDL, the uncertainty associated with that value increases, thus the 5x RDL rule is employed when evaluating duplicate acceptance limits.
See Calculating RPD for additional information.