The water distribution network (WDN), a key component of the water supply system, is a critical urban infrastructure distributing potable water to society. With network pipes as a key component and design variable, the single-objective design of WDN is a familiar non-deterministic polynomial-time hard problem. The present study proposes two variants of genetic algorithm (GA) with and without elitism operator, Simple_GA and Elitist_GA, for designing a real-scale WDN. Their optimization framework is structured with three main operators, truncation method of selection mechanism, single-point crossover, and bitwise mutation. Demonstrating the computational performance of GA variants, the study also highlights the critical nodes, which are of utmost importance for carrying out the performance study. The Balerma irrigation network (BIN), a gravity-fed large-scale benchmark problem with 454 network pipes, is considered for this purpose. On optimally designing BIN, the Elitist_GA model manifests the solution precision, yielding the hydraulically feasible optimal solution of 2,061,701 units over the Simple_GA model resulting in 2,089,535 units. Further, comparing the hydraulic conditions, the pressure head at most BIN nodes is above 30 m, with the maximum head falling around 70 m (approximately). Notably, out of 443 demand nodes, only the 16th and 39th nodes corresponding to the optimal solutions of Simple_GA and Elitist_GA, respectively, are on the verge of the minimum design pressure head, 20 m, and are ascertained as critical nodes. Overall, from the computational results, the Elitist_GA demonstrates better solution precision than Simple_GA, exhibiting better convergence properties for the same computational load of maximum allowable function evaluations.