Andingmen　station　with　non-enclosed　screen　doors　and　its　adjacent　section　in　Beijing　Metro　Line　2　was　taken　as　analysis　object.　The　dynamic　mesh　model　in　the　CFD　software　Fluent14.5　was　adopted　to　simulate　the　time-dependent　variations　of　air　flow　characteristics　during　train　operation　in　subway　station　and　adjacent　section　under　different　setting　modes　and　locations　of　air　shafts　as　well　as　different　lengths　of　section.　The　influence　of　air　shaft　setting　modes,　location　of　air　shaft　and　the　length　of　section　on　the　exhaust　rate　of　air　shaft　and　the　suction　rate　of　ground　exits　was　compared　and　analyzed　to　optimize　the　air　shaft　setting　in　subway　tunnel.　The　results　show　that　the　dynamic　mesh　model　in　the　CFD　software　for　analyzing　piston　wind　is　proved　to　be　effective.　Compared　with　double　air　shaft　setting　mode,　with　single　air　shaft　setting　mode　which　only　sets　piston　air　shaft　in　the　entrance　of　the　platform,　properly　increasing　the　distance　between　air　shaft　and　the　entrance　of　the　platform,　and　one　more　air　shaft　when　section　tunnel　is　longer,　the　piston　effect　produced　by　subway　train　in　tunnel　can　be　better　utilized　to　improve　the　air　environment　in　subway　station　and　thus　to　reduce　the　energy　consumption　of　subway　HVAC　system.　©　2015,　Chinese　Academy　of　Railway　Sciences.　All　right　reserved.