Network artificial chemistry (NAC) uses a mathematical graph to emulate molecular interaction in a solvent. To emulate molecules' movement in a three-dimensional space, rewiring rules for NAC graphs’ edges must be designed to enable the edges to imitate the relations between molecules or atomic clusters. Our research formulated the ‘network energy’ representing this constraint and rewired the NAC graph to minimize the required energy. Experimental results for the NAC rewiring are compared with a hard-sphere random walk simulation.