Representative year selection

Dear SWMMers:

Can somebody suggest a literature reference for representative year selection for analysis of combined sewer overflow? Your experiences in this area would also be of interest.

Thanks,

Mitch Heineman
HeinemanM@CDM.COM

Mitch,

Try the book by Adams and Papa (2000): Urban Stormwater Management Planning with analytical probabilistic Models (John Wiley and Sons). I guess a good way to go about this is to look at rainfall characteristics, do a statistical analysis on these and try to come up with years with average characteristics and, finally, run these 2 or 3 years with continuous simulation. The book mentioned has good discussion on rainfall analysis and provides equations that enable you to use the rainfall characteristics to obtain information about CSO, detention volumes and spills. It is derived from the approach started out at the end of the 1970s by Charles Howard (in B.C.) and that has been extended by Dr. Adams at U. of Toronto for the last 20 years or so.

Hope this helps.

Gilles Rivard, ing. M. Sc.
grivaquapraxis@VIDEOTRON.CA

Hi Mitch -

Not exactly the same thing, but maybe close. This was for an irrigation project in Turkey back in the 60s - our crew did not have access to a computer but had a bunch of alternative water supply sources and alternative irrigation service areas to sort out. We derived a "typical year" of monthly streamflow values, based on the geometric mean rather than the arithmetic mean of our meager data. The arithmetic mean was too distorted by outliers. Then we also had a low water supply year case to test with. Each alternative was described with a (real paper) spreadsheet with water allocations, costs, etc.

The computerized approach outlined by Giles and followed up with several years of simulation is a good one. Applying the Turkey case would suggest using geometric means for monthly precip. The several years of simulation should include dry, typical, and wet years for a comprehensive presentation of project behavior under different conditions. This approach assumes that over-year storage is not a significant aspect of the system.

Ron Kilmartin
ronkilmartin@attbi.com

Check with King County, Washington (was Surface Water Management and now is Bureau of Land Management). They developed a reduced set of 8 years, I think, of rainfall data that contained what they felt was a statistically representative selection of storms from a 50 year record in order to analyze stormwater systems there. After developing the representative years, they also reprogrammed a stripped down version of HSPF and canned it so they could actually make developers size detention vaults with it. They called the program KCRTS for King County Runoff Time Series. I don't have a specific reference, but I believe Dr. Steve Burges and Dennis Lettenmeier of University of Washington assisted King County with this.

http://dnr.metrokc.gov/wlr/dss/kcrts.htm

I don't have a reference; however, I can provide an approache I have employed on a number of occasions for CSO studies in the past where we we interested in finding a year that best reflects average conditions in the historical record. This does not identify a theoretical average year, if that is what you are after.

Use the Rain and Stats block to analyze your availabe years of say hourly data. From this analsis you can begin to select years with the representative characteristics in the following order: annual precip., annual number of events, monthly volumes, monthly # of events, and any other characteristics that you may be interested in. I have included monthly maximum intensities as well. We've done this kind of analysis on 30-years of data in just a few hours and ended up choosing an historical year as a an "average year". It's not perfect, but I have found it a useful strategy depending on the objectives.

Daniel P. Davis, P.E.
dpdavis@brwncald.com