The　pancake　graph　is　an　interconnection　network　that　plays　a　vital　role　in　designing　parallel　and　distributed　systems.　Due　to　the　unavoidable　occurrence　of　edge　faults　in　large-scale　networks　and　the　wide　application　of　path　and　cycle　structures,　it　is　essential　and　practical　to　explore　the　embedding　of　Hamiltonian　paths　and　cycles　in　faulty　networks.　However,　existing　fault　models　ignore　practical　distributions　of　faulty　edges　so　that　only　linear　edge　faults　can　be　tolerated.　This　paper　introduces　a　powerful　fault　model　named　the　partitioned　fault　model.　Based　on　this　model,　we　study　the　existence　of　Hamiltonian　paths　and　cycles　on　pancake　graphs　with　large-scale　faulty　edges　for　the　first　time.　We　show　that　the　n-dimensional　pancake　graph　Pn　admits　a　Hamiltonian　path　between　any　two　vertices,　avoiding　∑i=4n((i-4)((i-2)!-2)-1)+1　partition-edge　faults　for　4　≤　n≤　7,　and　avoiding　∑i=8n((i-1)!2-1)+399　faulty　partition-edges　for　n≥　8.　Moreover,　we　prove　that　Pn　admits　a　Hamiltonian　cycle,　avoiding　∑i=4n((i-4)((i-2)!-2)-1)+2　faulty　partition-edges　for　4　≤　n≤　7,　and　avoiding　∑i=8n((i-1)!2-1)+400　partition-edge　faults　for　n≥　8.　The　comparison　results　show　that　our　results　are　a　large-scale　enhancement　of　existing　results.　©　2022,　The　Author(s),　under　exclusive　license　to　Springer　Nature　Singapore　Pte　Ltd.