Morning　commuting　trips　commonly　involve　a　number　of　transportation　modes　and　departure　times.　This　study　extends　the　standard　bottleneck　model　by　considering　both　regular　and　shared　autonomous　vehicles　with　and　without　parking　space　constraint,　taking　into　account　the　travel　time-dependent　fuel　cost　that　is　associated　with　each　vehicle.　In　this　study　the　parking　space　constraint　has　no　effect　on　shared　autonomous　vehicles　and　only　commuters　who　share　autonomous　vehicles　exhibit　ridesharing　behavior,　differing　from　those　utilizing　regular　vehicles.　The　dynamic　departure　patterns　and　endogenous　penetration　rates　associated　with　shared　autonomous　vehicles　are　determined　with　respect　to　parking　capacity.　Analytical　results　not　only　provide　several　important　propositions,　but　also　include　the　optimal　solutions　for　parking　capacity　and　ridesharing　occupancy　from　the　perspective　of　the　system　optimum　in　terms　of　various　indicators,　under　which　parking　capacity　is　a　binding　constraint,　and　the　first　shared　autonomous　vehicle　arrives　at　the　bottleneck　as　the　last　regular　vehicle　leaves.　This　work　is　expected　not　only　to　motivate　related　research,　but　also　to　promote　the　development　of　new　strategies　and　methods　for　allocating　a　limited　number　of　parking　spaces　and　regulating　the　travel　behavior　of　commuters　in　urban　areas.　©　2019　Elsevier　Ltd