Temporal trends in green roof detention performance ... Green roofs are a type of ... max. Where R is runoff in mm, ET is evapotranspiration in mm, P is rainfall in.
Temporal trends in green roof detention performance as determined through reservoir routing parameters Simon De-Ville1, Virginia Stovin1 1.
Green Roof Research at Sheffield
Green Roof Ageing
Roof Test Bed Facilities, Sir Robert Hadfield Building, University of Sheffield.
Green roofs are a type of source control Sustainable Drainage System (SuDS) capable of reducing and detaining stormwater. The Hadfield Test Beds have been in place for 6 years. During this time they have generated a comprehensive record of rainfall and runoff. The first 4 years of this record are analysed here.
Green roof substrates are subject to a number of processes that are likely to result in changes to their hydrological performance as they age. Key processes include: root system development organic matter turnover substrate consolidation something else
Modelling Detention
Modelling Hydrological Processes Detention
Detention occurs as a result of the water having to pass through the layers of the roof before becoming runoff, hence detaining the flow. e.g. slowed by passing through the soil substrate. This can be modelled using simple reservoir routing techniques:
Net Rainfall Observed Runoff Modelled Runoff
1.2 1.0
{
Where R is runoff in mm, ET is evapotranspiration in mm, P is rainfall in mm and S is storage in mm. The retention proportion is then:
RtProp = (P-R)/P
TB8
TB9
0.05 0 0.10
medain k
TB6
Sedum
medain k medain k
0.10
TB7
0
Y1 to Y4 Median k 7
8
9
10 11 Time (Hours)
net-rainfall profile from observed rainfall and runoff data. Bottom: Observed and Modelled runoff for a moderate rainfall event. The model has an Rt2 value of 0.967.
12
13
Implications of Time-Varying k
Evapotranspiration, ET
TB9 - Yearly
2
Rainfall Net Rainfall Year 1 Runoff Year 4 Runoff
1.8
Storage, S
Storage Capacity, Smax
Excess Rainfall to Drainage Layer Routing
Runoff, R
1.6 1.4 1.2 1 0.8 0.6 0.4 0.2
Figure 2: Systematic diagram
of hydrological processess within a green roof system.
0
0
20
40
60
Time (minutes)
Simon De-Ville | The University of Sheffield | June 2014 | sdvresearch.co.uk
80
100
A
1 2 3 4 Year of Study
112%
14
A
0.05 A
0.2
Rainfall, Runoff (mm/min)
Rt=
TB5
0
0.096
Retention
0, St-1+Pt-ETt≤Smax Pt-(Smax-St-1)-ETt, St-1+Pt-ETt>Smax
TB4
0.05
0.4
Figure 3. Top: Generated
Rainfall, P
Retention is the process whereby the rainfall is actively retained and prevented from becoming runoff. This is modelled using a moisture flux method.
TB3
TB9 Median k, Y4
0
LECA
TB2
0.045
0.6
Which results in reduced detention performance
TB1
TB9 Median k, Y1
0.8
Leads to a rise in the value of k
Substrate Type SCS
HLS
1.4
Qoutt = k.hnt-1
Where Qin and Qout are the flow rates into and out of the substrate layer, measured in mm/min. h is the depth of water temporarily stored within the substrate, in mm. ∆t is the discretization time step. k and n are the reservoir routing parameters where k has the units mm(1-n)/min and n is dimensionless.
0.10
Rainfall, Net Rainfall, Runoff (mm)
The extensive green roof systems are configured with 3 vegetation and 3 substrate types, resulting in 9 total configurations. The three vegetation types are Sedum, Meadow Flower and an unvegetated control. The three different substrates are a Heather with Lavendar substrate (HLS), a Sedum Carpet substrate (SCS) - both of these are brick based - and the final substrate is based on a lightweight expanded clay aggregate (LECA).
A fall in organic content or substrate depth
Changes in k with Time
As detention 15 metrics can be Rainfall Net Rainfall influenced by Runoff 10 retention effects, the inital lossess of 5 each runoff profile were eliminated by substratcing them 0 0 2 4 6 8 10 12 14 from the total Time (Hours) rainfall. This results in a net-rainfall profile. Between 71 and 136 events were used to identify the best-fit k value for each test bed. For each of the runoff profiles a fixed value of n = 2 was used, k was then optimised using the lsqcurevefit function in Matlab. Goodness of model fit is indicated by the calculation of an Rt2 value. The runoff profiles and their associated k values were then categorised by study year.
Rainfall, Runoff (mm/5min)
Green Roof Configuration
ht = ht-1 + Qint.∆t - Qoutt.∆t
It is difficult to attribute changes in hydrological performance to a particular process individually. Therefore the reservoir routing parameter k represents a combination of all these effects.
A
1 2 3 4 Year of Study
A
A 1 2 3 4 Year of Study
Unvegetated Meadow Flower Vegetation Type
Figure 1: Current Green
Department of Civil & Structural Engineering, University of Sheffield
The vegetated test beds (TB1-6) see no statistically significant variation in yearly-median k and any observed variation does not appear to be systematically related to time. The unvegetated test beds (TB7-9) experience a large variation in the yearly-median value of k, with TB9 showing a steady year-on-year increase. These unvegetated beds have a statistically significant difference in k between year 1 and year 4 of the study (indicated by A-A). Any trends in median k for substrate type are harder to identify due to the dominant effects of vegetation.
Figure 4: Yearly
The increase in k can be seen to negatively impact on all 3 of the highlighted common detention metrics. Need something here that is quite conclusiony. Reiterate that this is a worst case example and that for those vegetated beds the differences would be much smaller. It is therfore preffereable to use a vegetated system to achieve continual levels of detention performance.
median value of k. Determined through optimisation of the hydrological model for each of the nine test bed configurations.
Peak Runoff
35%
Peak Attentuaiton
16%
Centroid Delay
----%
Figure 5: Runoff
response from an M30 60 minute 75% summer storm for Year 1 and Year 4 median value of k from TB9.