Smart grids with self-healing (SH) capability provide an important intelligent feature to help in fast correction actions in case of network faults. SH architecture consists of modern communication systems, smart equipment, and intelligent sensors. With the high cost of SH components (especially smart ring main unit (SRMU)), optimization is required to achieve optimum performance with minimum cost. This study presents a proposed methodology to determine the optimum number and locations of SRMUs in electricity distribution networks considering various cost issues. The disconnection cost of on-grid photovoltaic (PV) plants is taken into consideration as an important factor in determining the locations of the SRMUs. The nonlinear programming (NLP) optimization technique is used to determine the required number of SRMUs, considering the cost/benefit analysis (cost of upgrading MRMUs to SRMUs/benefit due to interruption time reduction), which is the most important factor from DISCOs’ perspective. The mixed integer linear programming (MILP) optimization technique is employed for selecting the optimal locations of the SRMUs considering the cost of losses, energy not supplied (ENS), and PV disconnection, which improves network operation cost. The methodology takes into consideration the cable failure rate and the interest rate. Moreover, the study introduces the Egyptian electrical distribution network and a pilot project for control centre development using SRMUs. The methodology is applied to a modified IEEE 37-node test feeder and a part of a specific district network in South Cairo consisting of 158 nodes; both systems include a number of PV distributed generation plants. Simulation results are presented to show the effectiveness of the proposed method.