I've studied the reliability ...

I've studied the reliability of municipal storage reservoirs using stochastic analysis (modelling user demand, fire demand and supply pipe failures) and have published a couple of papers detailing our results (can post them if anybody is interested). Here are a couple of our findings:

The number of failures per annum reduces exponentially with the tank capacity.

The reservoir failure risk is highly sensitive to the ratio of the weekly average demand to the inflow capacity, i.e. there is a great seasonal variation in reliability with the summer peak being the most critical time for failures.

We analysed a typical residential area where the reservoir capacity requirement was 52, 59 and 49 hours of annual average demand for the US, France and South Africa respectively. Based on an acceptable failure risk of 1 failure in 10 years under seasonal peak conditions, we found the required capacity to be 27 hours, which is substantially lower than those recommended by the design guidelines.

The reliability of a storage reservoir is not only a function of the size of the storage reservoir, but also of the capacity and configuration of the pipe system supplying it. For gravity systems, the optimal solution seems to be near a supply capacity of 1.3 times the seasonal peak demand with a reservoir capacity of 24 hours (but varies according to local conditions).

I hope this helps,

Kobus