This paper proposes a novel robust optimization (RO) approach along with a two-stage scenario-based stochastic programming to optimize a municipal water distribution system (WDS) under demand and rainfall uncertainties. Firstly, we have proposed a new multi-period mixed-integer linear programming (MILP) formulation of a municipal WDS. The goal is to find solutions that are both cost-effective and completely fulfill potable and non-potable demand in an integrated system. Furthermore, a novel RO approach is developed which attempts to adjust protection level in a column what we call “adjustable column-wise robust optimization”. The interesting point of the proposed RO approach is its linear structure and being computationally tractable. The efficiency of the proposed models are evaluated through a real case study of Mashhad. The acquired results reveal the proposed WDS model have dramatically reduced the total costs. Simultaneously, the RO approach has risen robustness besides realization demonstrates its better performance than deterministic one.

本文提出了一种新颖的鲁棒优化（RO）方法，以及基于情景的两阶段随机规划，以在需求和降雨不确定的情况下优化市政供水系统（WDS）。首先，我们提出了市政WDS的新的多周期混合整数线性规划（MILP）公式。目的是找到既经济高效又能完全满足集成系统中可饮用和不可饮用需求的解决方案。此外，开发了一种新颖的RO方法，该方法试图调整列中的保护级别，我们称之为“列可调整的鲁棒性优化”。提出的反渗透方法的有趣之处在于它的线性结构并且在计算上易于处理。通过对马什哈德（Mashhad）的真实案例研究评估了所提出模型的效率。获得的结果表明，提出的WDS模型已大大降低了总成本。同时，RO方法提高了鲁棒性，除了实现表明其性能优于确定性方法外，还提高了鲁棒性。