In the design of a wastewater treatment plant, engineers must estimate the future population that the plant will need to serve. When the population growth is uncertain, modular design can reduce up-front capital costs, operating costs and total expected present value costs. Models based on real option theory can provide insights to engineers/decision-makers that will not arise from standard net present value analysis. We consider a municipality faced with a plant expansion decision required to meet the demand of significant, uncertain growth. We present a model that optimises the size of the plant and its maximum modular size. The optimisation requires that multiple solutions be calculated at the decision time for the staged expansion. We utilise a moment matching technique to approximate the solution of the Asian-like option. The results show that as the uncertainty increases, a modular plant can save significant total costs for the municipality.