In　most　cases,　TBM　tunnels　are　constructed　in　jointed　rock　masses.　Rock　chip　properties　can　be　used　to　interpret　the　response　of　the　rock　mass　to　the　TBM　tunnelling.　Engineering-scale　simulations　can　help　us　to　explain　the　relationship　between　rock　chip　properties　and　rock　mass　conditions.　In　this　study,　a　GPU-based　DICE2D　programme　is　presented　to　meet　the　computation　requirements　of　the　engineering-scale　simulation.　The　smooth　joint　model　(SJM)　was　also　incorporated　into　DICE2D　via　an　improved　joint　side　checking　approach.　Using　the　SJM-embedded　DICE2D,　a　multiple-cutter　penetration　model　was　set　up.　The　combined　influence　of　joint　spacing　and　angle　on　the　rock　breaking　process　and　rock　chip　properties　were　analysed.　Further,　a　rock　chip　extraction　algorithm　was　proposed　to　quantify　the　shape　and　size　distribution.　The　results　showed　that　in　the　rock　masses　with　the　close　joint　spacing,　the　smaller　the　alpha　angle,　the　blockier　the　rock　chips　were.　As　the　joint　spacing　increase,　the　rock　chips　become　flat.　Both　the　joint　spacing　and　alpha　angle　affected　the　size　distribution.　The　overall　size　of　rock　chips　decreased　with　the　increasing　joint　spacing,　and　first　increased　with　the　increasing　alpha　angle　and　then　decreased　when　the　alpha　angle　was　over　60　degrees.　The　very　closely　jointed　rock　mass　at　the　alpha　angle　of　90　degrees　was　an　exception　because　rock　slabs　would　be　produced.　In　addition,　in　the　jointed　rock　mass　with　the　joint　spacing　less　than　250　mm　and　60　degrees-75　degrees　alpha　angle,　or　in　the　highly　jointed　rock　masses　with　90　degrees　alpha　angle,　significant　differences　between　the　forces　acting　on　the　different　cutters　were　revealed.