Continuous rainfall

The regression line in the following figures shows an average reduction in the discharge peaks caused by the LIDs.

The difference between the BRC II and BRC III at AK52 is not noticeable for the continuous rainfall for model setup A or model setup B, even for 80% disconnection or 100% disconnection as shown in figures 5.11 and 5.12.

The average discharge peaks can be reduced with 12% (y = 0.88x) and 22% at AK52 for re-spectively 80% BRC and 100% BRC with model setup A.

Figure 5.11: Simulations of the discharge peaks with BRC of 80% and 100% for two types, BRC II and BRC III, against simulations with no LIDs (scenario 0). The regression line repre-sents the mean discharge magnitude at AK52 after the LIDs are implemented in model setup A.

The average discharge peaks can be reduced with 14% (y = 0.86x) and 23 % at AK52 for respectively 80% BRC and 100% BRC with model setup B.

The slopes of the regression lines in model setup A and B are not significant different from each other for either 80% or 100% disconnection after using a t-test with significance level of 5%

and a sample size of 17.

Figure 5.12: Simulations of the discharge peaks with BRC of 80% and 100% for two types, BRC II and BRC III, against simulations with no LIDs (scenario 0). The regression line repre-sents the mean discharge magnitude at AK52 after the LIDs are implemented in model setup B.

The rain barrels (RB) are implemented in the same way as the BRCs. In model setup A, the average reduction in the discharge peaks are 22% for 100% and 11% for 80% RB for AK52 as shown in figure 5.13. Model setup B (see figure 5.14) reduce the average discharge peaks with 22% for 100% RB and 13% for 80% RB at AK52. There are no significant difference between the average reduction between model setups A and B.

The disconnection of downspouts is implemented with the re-routing option in SWMM (sub-section 2.4.8). In the setup of disconnection of roof area, the connected road area before im-plementation of LIDs is assumed the same after the imim-plementation. The total disconnection percentage is of that reason larger for model setup A than B, since model setup A initially has a higher disconnection rate than model setup B (see subsection 4.5)

In model setup A, the average discharge peaks are reduced with 23 % for 100% RD and 13%

for 80% RD at AK52.

Figure 5.13: The LID measures RD for 80% (upper left) and 100% (lower left) roof disconnec-tion at the Grefsen plateau. Simuladisconnec-tions with RB where 80% (upper right) and 100% (lower right) of the roof area area disconnected with rain barrels. The regression line shows the average reduction in the discharge peaks at AK52 with model setup A.

In model setup B, 100% RD gives an average reduction of 24 % on the discharge peaks at AK52. And 80% RD gives a mean reduction in the discharge peaks of 13%.

Figure 5.14: The LID measures RD for 80% (upper) and 100% (lower) roof disconnection at the Grefsen plateau. Simulations with RB where 80% (upper) and 100% (lower) of the disconnected roof area with rain barrels. The regression line shows the average reduction in the discharge peaks at AK52 with model setup B.

The reduction in the discharge peaks simulated with 100% roof disconnection at AK52 are not significantly different from each other. Also, 80% roof disconnection reveal that the average discharge reduction is not significant different for the different LIDs.

At Jupiterjordet, where all the sewer system is CSS, the average relative reduction in discharge peaks for the LIDs is higher compared to AK52. The wastewater from the householders is not included in the discharge. Therefore, the percentage average reduction is only for stormwater.

In figure 5.15, the average reduction in the discharge peaks are 43% with 80% BRC and 76%

with 100% BRC with model setup A. There is no difference between the BRCs in model setup A.

Figure 5.15: Simulations of the discharge peaks with BRC of 80% and 100% for two types, BRC II and BRC III, against simulations with no LIDs (scenario 0). The regression line repre-sents the mean discharge magnitude at Jupiterjordet after the LIDs are implemented in model setup A.

In model setup B, there is no significant difference for BRC II and BRC III (see figure 5.16).

For 100% BRC the average reduction is 57% and for 80% BRC the average reduction is 33%.

The average reduction with 100%BRC with model setup A is higher than what is simulated for model setup B. The same is observed with 80% BRC.

Figure 5.16: Simulations of the discharge peaks with BRC of 80% and 100% for two types, BRC II and BRC III, against simulations with no LIDs (scenario 0). The regression line repre-sents the mean discharge magnitude at Jupiterjordet after the LIDs are implemented in model setup B.

In figure 5.17, disconnection of 80% of the roof area with RB gives an average reduction in discharge peaks of 44% in model setup A. When all the roof is disconnected, 100% RB, the average discharge peaks is reduced with 78%. In model setup B an average reduction in the discharge peaks of 34% is observed for 80% RB and 67% for 100% RB (see figure 5.18).

The 80% RD gives a reduction in the average discharge peaks with 46% for model setup A (see figure 5.17). For 100% RD the regression line has a slope of 20, indicating a reduction of 80%. However, the coefficient of determination, R², to the regression line is very low. Similar is evident for 100% BRC and 100% RB with model setup A. These results must be interpreted with caution. R²describes the linear fit of the regression line. In model setup B, 33% reduction in the discharge peaks is observed for 80% RD and a 60% reduction for 100% RD as shown in figure 5.18.

Figure 5.17: The LID measures RD for 80% (upper left) and 100% (lower left) roof disconnec-tion at the Grefsen plateau. Simuladisconnec-tions with RB where 80% (upper right) and 100% (lower right) of the connected roof areas are disconnected with rain barrels. The regression line shows the average reduction in the discharge peaks for Jupiterjordet with model setup A.

The average reduction in the discharge after implementation of LIDs in model setup A is com-parably larger than for model setup B. There is no significant difference between the 100% RD or 100%RB with model setup A. The same is also observed for model setup B.

Figure 5.18: The LID measures RD for 80% (upper left) and 100% (lower left) roof disconnec-tion at the Grefsen plateau. Simuladisconnec-tions with RB where 80% (upper right) and 100% (lower right) of the connected roof areas are disconnected with rain barrels. The regression line shows the average reduction in the discharge peaks for Jupiterjordet with model setup B.