Water Journal : Water Journal April 2011
refereed paper technical features 140 APRIL 2011 water water supply options mwu with tank installation should reduce from 0.179kL/house/day to 0.136 kL/ house/day or 50 kL/house/annum. Figure 9: Single residential water consumption (srwc) relative to pre- and post-tank meter reads. Averages and 95% confidence intervals are shown. The next question to ask with regard to water demand is whether the performance of rainwater tanks is affected by changes in water demand. This may be answered statistically with a two-way analysis of variance of √mwu by tank installation and srwc. This analysis shows that tank installation (F=280, p<2.2e-16), srwc (F=317, p<2.2e-16), and their interaction (F=7.73, p=0.005) are all significant factors contributing to mwu. The interaction between tank installation and srwc indicates that there are linear effects of srwc on √mwu but with different slopes for the pre- and post-tank meter reads. Figure 10 shows these relationships, giving lines of best fit, 95% confidence bands and equations for pre-tank mwu by srwc, post-tank mwu by srwc, and the saving in mwu by srwc calculated as the difference between the pre- and post-tank lines. Figure 10 illustrates that tank installation gives a significant saving in mwu but that this generally reduces as community demand increases. High community demand generally corresponds with hot, dry weather, during which time rainwater tanks are more likely to be empty, so it is not surprising that average mwu among the rainwater tank group should rise towards the average consumption in the LGA when demand is high. Effect of tank-house connectivity Figure 11 shows mwu pre- and post-tank installation for the three most common rainwater tank connections: external (1-e); external, laundry and toilet (3-elt); and whole house (5-h). Once the expected reduction in mwu is subtracted due to the reduction in srwc over the corresponding period, water savings from these three connection types average 0.117, 0.075 and 0.294kL/house/day (43, 27 and 107kL/house/annum) respectively. T-tests show that these reductions in mwu with tank installation are significant for all connection types (T=-20.4, -8.8 and -11.6 respectively, p values are all <2.2e-16). Figure 11 shows that for these three connection options, 5-h connections result in greater water savings than the other connection types. Surprisingly, the 1-e group achieves higher water savings than the 3-elt group, which may be explained by the 1-e group having a significantly higher pre-tank water mwu (0.595 kL/house/day) than the 3-elt group (0.472 kL/house/day). Connection options may be related to differences in water-use habits, as external-only connections appear to be preferentially chosen by households with high water use, whereas external, laundry and toilet connections appear to be preferentially chosen by low water users. External use of tank water is presumed in all cases, as all tanks are fitted with a tap and it is a requirement of the rebate that tanks be available for external water use; however, the actual volumes of tank water used for each purpose is at the discretion of the occupants. The combined effect of tank installation, water demand and connectivity on mwu is analysed with a 3-way ANOVA. The results of this analysis show that connectivity is a strong predictor of √mwu (F=30.7, p<2.2e-16) even after the effects of tank installation (F=336, p<2.2e-16) and srwc (F=306, p<2.2e-16) are taken into account. In addition, the interaction between tank installation, srwc and connectivity is also significant (F=4.13, p=1.05e-5), so we can say that tank installation reduces mwu; however, the amount of this reduction during any given meter period is dependent on both connection type and srwc. These relationships are illustrated for the 1-e, 3-elt and 5-h groups in Figure 12. Water savings rise with srwc in the 1-e group (Figure 12 top), and fall as srwc rises in the 3-elt (Figure 12 middle) and 5-h (Figure 12 bottom) groups. This indicates that, while rainwater tanks connected for external-only use (1-e) generate relatively small water savings, they are less likely to fall empty during dry periods and so continue to contribute water to the dwelling. Whereas rainwater tanks that supply for the consistent Figure 10: Lines of best fit and 95% confidence range for mwu pre- and post-tank installation, and metered water savings, relative to the community's single residential water consumption (srwc). Figure 11: mwu pre- and post-tank for external only (1-e); external, laundry toilet (3-elt); and whole house (5-h) connection groups. Averages and 95% confidence intervals are shown. Change in metered water use (δmwu) is adjusted by the corresponding change in single residential water consumption (δsrwc) to give the reduction in metered water use (Δmwu) and the annual water saving.
Water Journal March 2011
Water Journal May 2011