Many real-world optimization problems involve inequality and/or equality constraints and are therefore posed as constrained optimization problems. A significant difficulty in multi-objective evolutionary optimization is the lack of a general explicit method for handling constraints. Typically a penalty function approach is used. This study assesses a potentially attractive approach for constraint handling based on the non-dominance concept. A case study based on the Canberra bulk water system is used to compare the performance of non-dominance based constraint handling and penalty function methods. It is shown that results from constraint handling based on the non-dominance concept outperform the best tuned penalty function results. Further improvement is needed to maintain infeasible solutions during the evolution thereby the search can be done both from the feasible and infeasible regions to approach the constrained boundary.