Skip to main content Skip to navigation

Improving the Viability of Roofwater Harvesting in Low-Income Countries

D. Brett Martinson

A thesis submitted in partial fulfilment of the requirements of Doctor of Philosophy in Engineering (2008)

Domestic roofwater harvesting (DRWH) provides an innovative solution to meeting water needs. There is renewed interest in the technology; however problems of implementation, concerns about water quality and health, and a reputation for high cost inhibit its wider take-up. This thesis is an investigation of two of the constraints, namely cost and water quality.

The performance of systems is discussed along with the losses at various points along the system. A number of tank-building techniques are investigated and economies of scale and cost ranges are developed. Lessons learned from a range of designs are then used to develop a set of
strategies whereby costs can be saved in DRWH tank construction. Two of these strategies are investigated theoretically:

  • The trade-offs of investment and servicing costs inherent in reduced quality construction are analysed using net present value analysis
  • Underground construction of cylindrical tanks is analytically modelled using a modified elastic structural theory which is verified with finite element analysis. More complex shapes are modelled using finite-element analysis alone

Several of the strategies developed are used to design tanks which are field tested. The strategies used are found to significantly reduce initial cost, however several designs require a discount rate to be introduced before they are viable long-term.

The second section deals with the quality of harvested roofwater, initially looking at the overall system, health risk pathways are traced and an overall picture of changes that take place over the course of these pathways is built-up. Treatment of incoming water from the roof is then
investigated, particularly concentrating on quantifying the first-flush phenomenon which is exploited in devices that divert the first “dirty” part of a rainstorm away from a store before allowing the “clean” part into the store. The effect is found to be slower than expected and most devices are too small and reset too quickly to e very effective. A rational method for sizing first-flush device based on desired material removal is developed using a mass-balance approach.

Full Document (355pp, 4.1Mb) (PDF Document)