Related Substances Method Validation

For validation of a HPLC Related Substances method, is it acceptable to use a batch for the validation that contains an unknown impurity above the specification level - or from a compliance perspective should the batch meet all specification limits?

Here you have been validating the analytical method for related substances… That means you have to prove that your method is capable for the identification of minute quantities of impurities in the analyte.
Selection of sample for AMV, which is having higher content of any impurity is not correct.
You have to consider compliance batches only…

Could the identification of minute quantities be shown by establishing LOD and LOQ?

Yes, You have to develop your test method, up to that level…