Power system grids are experiencing increasing photovoltaic (PV) installations, especially in the low voltage distribution grid. Reverse power flow caused by local power generation (such as from high penetration PV) brings several challenges to the grid, such as overvoltage. These challenges limit PV installation/operation in a distribution system, which affects owner revenue and reduces the positive environmental impacts. This paper provides a framework to estimate PV hosting capacity of distribution grids. In this paper, a Pacific Northwest National Laboratory taxonomy feeder is analyzed using Monte Carlo simulations. The PV penetration was varied from 10% to 100% in 10% steps, and for each penetration level 25 unique scenarios were randomly generated (where each scenario varies the random location of PV on the feeder). To reduce the total simulation time, a parallel algorithm was developed using Message Passing Interface (MPI) to split the scenarios across 25 processing elements. Power flow was run for the highest irradiance day and lowest load day of 2015 in a one-minute timestep for each scenario. The frequency of overvoltage occurences increased with increasing PV penetration level in both days, however overvoltage only occurred after 50% PV penetration for the low load day. The peak voltage magnitude and voltage mismatch between phases was higher for 50% - 80% penetration level, which is due to the random distribution of PV across phases.