Mesoscale　structures　such　as　clusters　are　common　in　gas-solid　two-phase　flows.　These　mesoscale　heterogeneous　structures　directly　affect　the　gas-solid　flow　characteristics　and　gas-solid　contact　efficiency,　and　then　affect　the　gas-solid　interphase　heat　transfer,　mass　transfer　and　chemical　reaction　process.　In　the　coarse　grid　method　which　is　more　suitable　for　industrial　large-scale　gas-solid　heat　transfer　simulation,　there　is　a　lack　of　heterogeneous　heat　transfer　model　with　high　accuracy,　simple　and　easy　to　use,　and　can　consider　the　influence　of　mesoscale　non-uniform　structure.　Computational　fluid　dynamics-discrete　element　method　(CFD-DEM)　is　used　to　study　the　interphase　heat　transfer　of　gas-solid　two-phase　flow.　In　order　to　ensure　the　continuous　heat　transfer　between　gas　and　solid,　two　methods　to　maintain　the　temperature　difference　between　gas　and　solid　during　the　heat　transfer　procedure　are　adopted,　and　the　advantages　and　disadvantages　of　the　two　methods　are　discussed.　Method　1:　add　a　heat　source　term　to　the　gas　phase　energy　equation;　Method　2:　reset　the　gas　phase　temperature　at　intervals　and　free　heat　transfer　between　the　gas-solid　two　phases　after　resetting　the　temperature.　The　solid-phase　temperature　remains　unchanged　in　both　methods.　The　results　show　that　the　local　gas-solid　heat　transfer　per　unit　volume　at　the　cluster　interface　is　the　largest.　The　ratio　of　local　gas-solid　heat　transfer　per　unit　volume　to　the　total　gas-solid　heat　transfer　per　unit　volume　at　the　dilute　phase　and　the　interface　of　the　clusters　in　the　testing　temperature　method　is　greater　than　that　of　the　heat　source　term　method,　while　the　ratio　of　local　gas-solid　heat　transfer　per　unit　volume　to　the　total　gas-solid　heat　transfer　per　unit　volume　at　the　dense　phase　is　less　than　that　of　the　heat　source　term　method.　By　filtering　the　CFD-DEM　calculation　data,　a　two　parameter　(filtered　solid　volume　fraction　and　filter　size)　heat　transfer　coefficient　correction　factor　model　is　constructed　for　the　reset　temperature　method.　The　performance　of　the　model　is　evaluated　through　a　priori　analysis.　The　results　show　that　the　proposed　model　is　better　than　the　existing　two　parameter　model　in　the　literature　when　the　filter　size　in　the　range　of　5　to　40　times　the　particle　diameter.　©　2022　Chemical　Industry　Press.　All　rights　reserved.