An increase in people’s awareness of climate change and its ensuing consequence of increased power failures have given rise to discussions on ways to enhance energy reliability, which includes transitioning to microgrid. With the vast majority of grid system still being conventional, it is more practical to build microgrids based on existing ones. This thesis proposes a specific algorithm to transform existing community scale power grid system to microgrids aiming at enhancing energy reliability and resiliency, and a real-life disadvantaged community is modeled and analyzed upon with added Distributed Energy Resource (DER) and Energy Storage System (ESS). The physical power flow model of the real-life disadvantaged community that is applicable to a mixed residential/commercial/industrial neighborhood was first developed on simulation software OpenDSS. Before applying the baseline load separately generated by DERopt tool, E3 RESOLVE model was used to verify its accuracy. Different levels of electric vehicle (EV) charging load were then estimated and added on top of the baseline community scale energy model designed to reflect on current trends. A revised multilevel Graph-Partitioning algorithm combined with Mixed-integer linear programming (MILP) was then developed and implemented to optimally separate the existing community grid connections into different islands that yield best possible static balance between local power supply and demand with added DER/ESS. Results shows that under various penetration of EV adoption and DER/ESS our algorithm works to provide the most reliable, resilient energy supply to the community.