<p>This study examines the optimization of single-phase reactive power compensation (RPC) in low-voltage networks. We use the particle swarm optimization (PSO) algorithm to analyze three optimization approaches: (i) using reactive-power sensitivity factors and normalized voltage, (ii) using only reactive-power sensitivity factors, and (iii) without using sensitivity factors. In the first approach, potential RPC nodes are identified using both reactive-power sensitivity factors and normalized voltage, while in the second approach, only reactive-power sensitivity factors are considered. The PSO algorithm then selects nodes and capacities of the necessary devices from the identified potential nodes. In the third approach, the PSO algorithm directly determines the connection points and capacities of compensation devices. In all cases, the same objective function-accounting for costs and voltage deviations-was applied. The proposed RPC optimization approaches were tested using a MATLAB model of an electrical network section, with the optimization problem and the PSO algorithm implemented in Python. Based on the results obtained, the third approach is the most effective for the network under consideration, reducing active power losses by 12% and improving voltage levels as well as the zero-sequence voltage unbalance factor at network nodes.</p>