PFC is powerful for exploring fracture coalescence, but reviewers rightly challenge models that only reproduce a final failure shape.
Our calibration ladder:
1. Unconfined and triaxial runs on synthetic assemblies
2. Direct tension and Brazilian disc proxies for tensile response
3. Jointed samples with mapped persistence ranges
4. Validation against AE or micro-seismic trends where available
Document every downgrade: if you simplify fracture sets for runtime, state which mechanisms you expect to under-predict. Pair PFC insights with continuum checks so design decisions are not made on a single code.