Hydrologic Modelling for Green Roof Systems 1

[bperry211]

I am looking into the use of a green roof system for an upcoming project. I realize there are a number of variables involved, but can anyone recommend the best way to model the use of sucha system. In particular, it would be important to model the green roof in combination with other conventional designs - i.e. a 'cookie-cutter' detention basin.
Any ideas are much appreciated.

 

[sam74]

I've worked on one project in the GSMNP where we were required to design to "green" standards. Can you elaborate on exactly what your green roof standards are? We were told this by the architects on that building also. We just assumed the CN for the building was the same as the surrounding forest and included a detention pond to account for parking area. Those green standards seemed to be pretty vague.

 

[bltseattle]

There are several ways this could be done.

Background: Because of climatic variability, such as the conditions between storms that allow a green roof to "recharge, green roof benefits vary geographically, so the modeling approach used in one region may not be accurate in a different region. There are no commercially available programs to model green roof hydrology. Modeling a green roof as a conventional soil doesn't make sense when you consider that green roofs are (a) shallow and underlain by impermeable layer (roof),(b) media is highly absorptive yet highly drained (what CN to use?), and (c) there is often a significant delay due to lateral flow through a drainage layer (what Tc to use?). I know this because, after painstakingly researching this subject, my firm has invested in developing a green roof hydrologic model. The Excel-based tool, which analyzes 24-hr design storms on a site that has a green roof, has been developed, was presented to the City of Seattle, was calibrated with the limited green roof data that is available, and has been peer-reviewed by a national green roof expert. We are currently collecting runoff data from five test plots in Seattle to refine the calibration of the tool. I can provide you articles outlining our modeling approach and monitoring program.

Options:
1. If you are using an Opti-grun or Roofscapes green roof, Roofscapes (Philadelphia) can apply the Opti-grun model developed from German runoff data. My engineering advice is to make sure your climate is compatible, however.
2. Hire us! Sorry for the promotion, but since I've spent 100s of hours developing this tool, and there is not a publicly available substitute, you can benefit from my long hours of R&D and expertise.
3. Borrow someone elses approach. For example, the City of Portland Bureau of Environmental Services allows green roofs to be modeled using SCS/SBUH methods, if you use the agency-specified curve number. You just have to convince your reviewing agency (and yourself) that it makes sense.
4. Use academic results. Research led by Prof Bill Hunt at NC State BAE has led to development of CNs and Rational C-factors for their green roof configuration in their climate. They have useful publications available on-line.

I am happy to discuss these options with you further either here on the forum or via the contact info below.

Brian Taylor
Magnusson Klemencic Associates
206-292-1200

 

[psmart]

Some degree of modeling may be possible with standard runoff models. Similar situations occur when modeling playing fields and other artificial situations. A CN value can be estimated by taking the SCS equation for the maximum possible retention:

S = 1000 / CN - 10 (S in inches)

And rearranging to solve for CN:

CN = 1000 / (S+10)

If you can determine the total voids in your soil (S), you can estimate a CN value. You can adjust for especially wet or dry soils by changing the Antecedent Moisture Condition, which is adjustable in many SCS/NRCS models such as TR-20 or HydroCAD.

Depending on the layout of your roof, Tc values are likely to be fairly short, but you could use standard travel-time calculations to estimate the Tc.

If there are ponding effects (roof storage), you can use conventional pond routing techniques.

The most difficult behavior to estimate would be the discharge from any underdrains. But if this is subject to a significant time delay (relative to the rainfall duration), it may not effect the peak discharge from the site. Whether you need to include this effect will depend on the exact objectives of your study.

 

[bltseattle]

Psmart is correct that you could compute a theoretical CN based on soil storage, it's a starting point, BUT:

* How do you account for antecedent moisture?

* Consider that SCS/NRCS engineers have said "It is sometimes inferred that the parameter S... is a physical property of the site like a soil moisture storage parameter...This has not been shown with any certainty. The parameter S (or CN) is a model variable and is only a constant for a particular storm." (reference: Runoff Curve Number Method: Beyond the Handbook

http://www.wcc.nrcs.usda.gov/hydro/hydro-techref-h&h-curve.html)

* How do you know that the 0.2S abstraction inherent in CN methods is relevant to green roofs (it's not!)

* How would you compute the time of concentration for a green roof? (- did you know that the time of concentration for a conventional CN analysis is not really about time, but it is derived from the slope of the storage:discharge curve for a particular basin, per the original SCS documents [Reference: SCS Hydrology Tech Note 2 Chapter 1 "The Modified Att-Kin Routing Model"]. I've performed detailed seepage analysis then back-calculated the corresponding Tc and shown it is can be much longer than on a conventional roof when an aggregate drainage layer is used for the green roof.)

* How do you account for flow time (and storage) in a lateral drainage layer?

I only bring these issues up because it is not clear or obvious how to adjust a CN method for green roofs, particularly if you are trying to estimate both runoff peaks and volumes accurately. If you must perform a CN analysis for the green roof consider that you might use different CNs for volume and rate analyses, and you must somehow account for the anticipated antecendent moisture.

My opinion is that the standard methods of analysis/models are ill-suited to account for BOTH volume and rate estimation for green roofs, and to analyze those aspects separately is not conducive to detention design based on level pool routing for a typical site, where "conventional" areas of the site would be analyzed using CN methods yet green roof hydrologic response could influence the detention sizing.

As a quick check, you might consider the extent of green roof coverage is on the site. If it is less than say 20%, it may not warrant extensive modeling as the rest of the site will likely drive any detention sizing.

BLT

 

[psmart]

These are legitimate concerns about the SCS/NRCS methodolgy in general, and not just for roof systems. But the broad use and acceptance of the SCS/NRCS methodology means that it will be applied to a wide range of situations. In addition, roof gardens aren't very different from "rain gardens" and micro-ponds, which are being widely modeled with the SCS/NRCS method.

For accurate models, many variables must certainly be considered, including the initial abstraction ratio (Ia/S). This can be adjusted in many programs, as can the Antecedent Moisture Condition, which causes an adjustment in the CN value to account for the pre-storm moisture level.

Since most roof systems will have a relatively short Tc, the exact Tc may not be critical. For a short Tc (right down to zero) the runoff happens so quickly that the rainfall distibution becomes the dominant factor in determining the peak flow and hydrograph shape. In this situation, the choice of Unit Hydrograph also has a minimal effect.

This reduces the critical parameters to the CN and Ia/S ratio. The underdrain flow is certainly more complex, but unless this occurs very quickly, it won't effect the runoff hydrograph until well after the peak runoff has occured. It all depends on the exact question you're trying to answer.