Much attention has been paid to understanding the clustering mechanism of water adsorbed on carbonaceous adsorbents. Adsorbed water forms clusters around strong sites, such as functional groups and surface defects, and these clusters then coalesce if the strong sites are sufficiently close to each other. Simulations of water adsorption are notoriously time consuming because of the slow relaxation of the strongly-directional hydrogen bonds. Our objective in this paper is to gain a better insight into clustering and coalescence of water, without incurring large computing overheads. To this end we have chosen argon as an adsorbate, and a substrate that is a very weak adsorbent for argon. To mimic functional groups, the substrate surface is decorated with strongly adsorbing patches. The adsorbate forms nano-clusters with convex surfaces at pressures greater than the saturation vapour pressure. When these clusters are sufficiently close to each other, they coalescence to form larger fused clusters, and there is a decrease in the equilibrium pressure. The relationship between the radius of curvature of the developed nano-clusters and the equilibrium pressure follows the functional form of the Kelvin equation, but the energy parameterγvM is smaller than the bulk value, implying that the clusters have a smaller cohesive energy.