Revisiting the Dangers of Naive Approaches to Transient Optimization of Water Supply Systems

Water supply systems must meet a variety of completing requirements under a broad range of operations, evaluated by performance measures ranging from cost, to reliability, robustness and resilience, not to mention apparently strict (or at least highly constraining!) regulatory stipulations. The hydraulic conditions to which the system is subject include a variety of demand scenarios (generally high demands, or low demands, with or without fires, to complex mixes of these conditions). The network can be in a variety of states of operation and repair, with different pumps on or off, operational reservoirs at various levels, or pipes having various states of integrity and serviceability. There are many time-scales at play, ranging from variations in the price of power, varying service needs and vulnerabilities at different times of day and year, the on-going evolution of land-use patterns and water use habits, various special vulnerabilities and requirements (like special meetings or various special needs alerts), and the on-going adjustments due to technical advances in sensors, equipment or regulatory requirements. Hydraulically the system must contend with instantaneous concerns with both high and low pressures to the longer-term (multi-day) evaluation of water age and the spatial distribution of disinfection agents (or their by-products!) . In the background, there are vexing and unresolved issues such as what it really means to violate a pressure or flow requirement.

Given this, with almost every parameter having a set of measurement challenges and operational uncertainties, what exactly would it mean to have an “optimum” system? And optimization is usually premised on the concept that there are available resources to be redeployed is some more beneficial way, but what if these resources are in fact crucial for some other operational state hereto unconsidered by the optimization framework? How does one make difficult tradeoffs between, say, the inability of a system to immediately fight a fire with the long-term health risk of a low-level exposure to some water-borne contamination?

This paper argues that it might sometimes be possible to recognize conditions that are generally wasteful, but that the quest for a true optimum may be as elusive as that for the fountain of youth. This paper will explore where the tradeoffs in design and operation are most complex and where discussion and debate may be more helpful in the short-run than a preoccupation with mathematical quests for optimality.