Design storm for detention facility

[heartlandce]

We're working for a City to evaluate options for regional detention to mitigate overtopping of an existing detention facility. We have a dynamic rainfall-runoff model that we're using to route runoff determined using the TR-55 method. This is a coupled 1D/2D routing model, so any discharge not conveyed in the pipe network is conveyed as overland flow or stored on the surface. If we're evaluating overall system performance, we typically use a nested frequency rainfall distribution, which is essentially an SCS Type II distribution updated with current rainfall depths from NOAA Atlas 14. This works well to evaluate pipe capacity within basins with varying times of concentration.

However, this approach produces high rainfall intensities that can overwhelm the pipes upstream in the basins, causing flooding, which could then alter the hydrograph at the proposed detention pond location as the water that couldn't be handled by the upstream pipes is potentially delivered more slowly to the detention pond location. I'm concerned that this approach could underestimate the volume required. I'm wondering if we need to look at a more gradual rainfall distribution, where the rainfall occurs at lower intensities over a longer period of time.

Am I overthinking this?

 

[cvg]

you might want to look at a variety of storms, 6-hour, 24-hour, 72-hour and NRCS even requires a 10-day storm be analyzed for dams. I would not expect the storm drains in the watershed upstream of your detention basin to handle 100-year peak runoff, they are usually not designed to handle that.

 

[LincolnPE]

If you are using SCS in an urban setting concerning detention volumes, make sure you use the weighted-volume approach and not area-weighting of CN as in TR-55.

 

[heartlandce]

Thanks for the info, folks. I've seen a lot of discussion on the boards for volume weighting through my search, but haven't heard much mention of it otherwise. In reading your responses, LincolnPE, and the New Jersey stormwater BMP manual, it looks like the difference is primarily related to the initial abstraction term and is most pronounced for the water quality volume, where the rainfall is 4 inches or less. Our 100-year rainfall depths are a lot more than 4 inches (more like 7-8), but it looks like something I should check into more.

 

[beej67]

LincolnPE brings up volume weighting v area weighting quite a bit around here, it's one of his pet issues. There can be advantages to doing it that way, but area weighting is very common in many areas of the country. If I tried to do volume weighting on a design in my region, the reviewers would kick it back to me and refuse to accept it.

To your actual question, what storm to use varies by region. In many areas, I've seen 25 year storm used for pipe sizing and 100 year storm used for detention sizing. Often areas with that sort of split criteria will demand that you produce some sort of documented proof that the 100 year rainfall makes it to the pond somehow (overland flow in gutters etc) if it exceeds the pipe capacity. That should be relatively easy with the analysis you're doing. If it jumps the curb and bypasses your pond for certain storms then your pond assumptions aren't valid.

Some places just require everything to be sized for 100 year. In South Florida, ponds are only sized for 25 year, because anything past the 25 year storm (= cat3 or bigger hurricane) they've evacuated and are watching their sites on TV from Valdosta. In Florida DOT projects, you have to run a storm duration sensitivity analysis on your site, to determine which duration storm is critical, then design your pond around that. So two different ponds in the same region both discharging to FDOT right of way might be designed for different duration storm distributions.



Hydrology, Drainage Analysis, Flood Studies, and Complex Stormwater Litigation for Atlanta and the South East -

http://www.campbellcivil.com

 

[mes11]

How does volume weighting provide a "better" or more accurate design. Aren't you using the same curve numbers either way you do it?