Smart grids are complex power networks that employ bi-directional communication amongst distributed energy resources (DERs), electrical grid customers, and a central control system that optimizes power generation while ensuring overall system safety and normal power operation. The self-regulating capability of smart grids can lead to improved power quality, optimal exploitation of renewable energy resources, network self-repairing in the event of outages, and an opportunity for customers to manage their electricity usage for reduced costs. In this work, an advanced global optimization search tool has been integrated with simulation models of a power distribution grid, traditional and renewable power sources, and distributed energy storage devices, to identify the optimal operation of the power distribution system. The newly-introduced, metamodel-based global optimization search tool was based on the Space Exploration and Unimodal Region Elimination (SEUMRE) algorithm for solving complex global design optimization problems that have black-box objective/constraint functions and which require extensive computation. The power distribution grid was modeled using the extended open-source software developed by the US DOE Pacific Northwest National Laboratory (PNNL), GRIDLAB-D [1]. The grid model is a low voltage, single distribution smart-grid, consisting of a distribution network based on the IEEE 13 test grid, with its original loads and transformer/substation. A single wind turbine generator is located in the IEEE 13 grid as a distributed renewable energy source and household heat-pumps are used as distributed energy storage devices. The design objective is to minimize the transmission/distribution power loss for a given set of distribution bus loads and wind power generation, while maintaining the distributed energy services using heat-pumps that are operated between given minimum and maximum temperature limits in virtual households. The thermal capacity of a customer’s home provides energy storage with variant heating levels, and household temperature monitoring feedback to the central control for effective and efficient power generation and distribution. The power loss was determined using a Forward-Backward Sweep (FBS) power flow calculation method for the three-phase steady state distribution grid bus over a time-series simulation. The optimization ensures minimum power loss, and efficient power generation and distribution to satisfy the power needs of customers.