EPA-SWMM: Subcatchment Width 1

[froude]

All,

I was wondering if anybody had any experience in calculating the "Subcatchment Width" parameter for use with EPA-SWMM. The SWMM manual suggests using the ratio of the subcatchment area to the overland flow length (or an average value, thereof). Does anybody know of any other methods?

Other sources have indicated that you can use the equation:

k * SQRT(Area) where k is between 0.3 and 0.5.

Does anybody know where this equation comes from and where documentation on this equation can be found?

Any suggestions or guidance would be appreciated.

 

[RWF7437]

You may find an answer at this website:

http://www.swmm2000.com/

Despite many hours, and hundreds of dollars on books and manuals, SWMM5 remains a mystery to me. If you find a plain english explanation of how SWMM works its magic, please share it with us.

 

[SMIAH]

The width W is a parameter that, conceptually, is associated with the response time of the basin.

When W is small, the flow path is long and the response will be slow (attenuating peak flow).

The more W is increased, the response will be quicker and the flow rate will potentially increase (less infiltration).

There are several possible approaches to determine the parameter W. Ideally, if flow measurements are available, one can use W to calibrate the model. However, usually those measurements are not available (e.g. a network that does not exist yet), W must be determined by considering the shape and characteristics of the sub-basins, with no possibility to adjust this parameter. It is therefore useful in this context to examine the sensitivity of this parameter.

In the software, the parameters W, the slope (S) and Manning's n are grouped as shown on the attached file.

Therefore, we could theoretically change S or n to obtain similar results as a change of W. Since the slope and n are determined more precisely, W is uses as a calibration parameter.

 

[SMIAH]

Possible definition of the parameter [see attached].

 

[RWF7437]

First, apologies to Froude for horning in on his thread but the subject interests me as much as it appears to interest him.

To SMIAH:

Thank you for your responses which have edged me a little closer to understanding SWMM. May I ask a few questions of you ?

1. You reference “Wisner, Kassam and Cheung (1983)”. I googled this reference but with few results. Is this a book, a Journal article ,or something else ? I could not find it for free or at Amazon.com.
2. In the equation Q = 1.49/n*(d-dp)^5/3(S^1/2) may we assume that:
Q = flow in ft^3/sec
dp=depth of infiltration losses ( feet)
d=depth of rainfall minus evaporation (feet)
w= characteristic width of watershed (feet)
S= slope of the hydraulic grade line (ft/foot).?

The units are important because this equation would be slightly different in metric.

3. If S is not the slope of the HGL what is it the slope of and how would we find it ?
4. Is the equation simply a form of Manning’s equation ?
5. If so does this imply that SWMM models runoff by assuming the entire basin is a very wide, shallow open channel with a rectangular cross section and a uniform slope and roughness?
6. Why do you say “ Since the slope and n are determined more precisely...” I don’t know of any way to determine n “precisely” outside of a hydraulics laboratory. In most real world situations n is an educated guess at best.
7. Similarly, how can we determine precisely the value of S. The slopes in any real basin vary considerably. There are many things S could be including the slope of the main watercourse, the average slope of all the land surfaces within the basin or even the often used slope of the main watercourse between the 10% and 85 % points. Or is it something else ?
8. Which way is the flow of the main watercourse in the sketch of a typical sub-basin you posted ?

Any help would be appreciated and thanks again for your comments.

 

[froude]

the continued advice from everyone is much appreciated and hopefully it keeps coming! (especially since EPA-SWMM is so new to me). certainly no need to apologize either!

thanks for the continued information.

 

[SMIAH]

RWF7347.

I'm gonna check for documentation about this parameter and post later. You have good questions!

 

[RWF7437]

Is this the reference ?

"A GIS-based model for urban flood inundation

"Journal of Hydrology, Volume 373, Issues 1-2, 30 June 2009, Pages 184-192
Jian Chen, Arleen A. Hill and Lensyl D. Urbano "

Do you have a copy ?
Can you post the pertinent excerpt ?
Does this have anything to do with SWMM or its derivatives ?

 

[SMIAH]

I'll try to answer some of the questions.
I'm not a fan of SWMMM so i never really went through it

1. You reference "Wisner, Kassam and Cheung (1983)". I googled this reference but with few results. Is this a book, a Journal article ,or something else ? I could not find it for free or at Amazon.com.

I can't find it neither. Probably mistaken.

2. In the equation Q = 1.49/n*(d-dp)^5/3(S^1/2) may we assume that:
Q = flow in ft^3/sec
dp=depth of infiltration losses ( feet)
d=depth of rainfall minus evaporation (feet)
w= characteristic width of watershed (feet)
S= slope of the hydraulic grade line (ft/foot).?

The units are important because this equation would be slightly different in metric.

Q = W (C/n)*(d-dp)^5/3*S^0.5

where
C = 1.0 SI and 1.49 SA
Q = subcatchment (or subarea) outflow, cfs,
W = subcatchment width, ft,
n = Manning’s roughness coefficient,
d = water depth, ft,
dp = depth of depression (retention) storage, ft,
s = slope, ft/ft.

For Q3,4,5 and 7 i'd refer to chapter 2 page 40:

http://eng.odu.edu/cee/resources/model/mbin/swmm/swmm_2.pdf

6. Why do you say " Since the slope and n are determined more precisely..." I don't know of any way to determine n "precisely" outside of a hydraulics laboratory. In most real world situations n is an educated guess at best.

True. I'd say that we can have a better guess for n than for W.

Wayne Hubert = Seems to be a good reference.

 

[RWF7437]

http://cce.oregonstate.edu/people/faculty/huber.html

 

[RWF7437]

Thanks again SMIAH,

It appears from the ODU ( Old Dominion University ?) reference you posted that SWMM routes the rainfall through an imaginary non-linear reservoir using the equation for Q you cited. Because I've never understood exactly what a "non-linear reservoir" is some confusion remains.

This also means that the slope in the equation must be the average slope of the main watercourse, I'm guessing.

The pesky parameter "w" is similarly confusing. If it is indeed a Parallel Measure of something we can't measure directly, like S, the question remains what is it a measure of ? If, as suggested, it is a measure of the rapidity of the runoff why would it be preferred to other such measures such as travel time, time of concentration, or lag time ? Although these things, Tc, Lag, etc., can be measured directly they seldom are.

It is also unclear why Area/length would be a good approximation of "w". Would it not be better to use calibration data from actual watersheds to make a first guess at "w"? SWMM has been in existence since 1969-70. Surely, by now, there is a large body of calibration data which could be used to make that first guess for the value of "w" closer to reality ?

I've been existence since 1937 and have never been required to learn or use SWMM. I think I'll plow blithely on without it for the next 50 years.

Thanks for your comments and I hope they help Froude to answer his original questions.

 

[froude]

I believe all this informaiton has been extremely useful and I thank both of you for it.

In some of the further research I have been doing, it appears that there is no concrete answer for the question I asked originally. The subcatchment width can be calculated in 1 of many ways:

1. Drainage Area / Overland Flow Length
2. k * SQRT(Area)
3. the list goes on...

But apparantly, the purpose of this variable is mainly as a calibration variable (as stated by SMIAH above). Other sites and forums have indicated that this variable can be manipulated in orer to match model flows to flows from physical flow gauges within the watershed. However, the problem arises when there are no gauges; i.e. you are blindly guessing. There are recommendations that say you should repeat method 1 (listed above) many times for a watershed and take the average of the values, but that still seems to arbitrary for me. I guess the hunt continues...

Thanks to all!

 

[SMIAH]

See the file attached.

This is an extract of a file written in another language (might be some errors in it).

 

[RWF7437]

http://www.waterbucket.ca/rm/index.asp?sid=48&id=412&type=single

 

[RWF7437]

To take but one example,

“Tests showed in 1970 that taking W as 1.7 x length of main sewer in a sub-basin for which we neglect the network gave good results.”

Although this is only an excerpt from an excerpt, one can only wonder what tests were conducted in 1970. Who conducted those tests. What kind of tests they were. How large the sub-basins were. What the results were. What they were compared to. How they were compared. How “good” the results were. What parameters were compared. And, how generally applicable the results are considered to be.

Some things seem lost in translation.

 

[SMIAH]

Yes that sentence is... not really helpful i guess.

I got this text in a SWMM course a few years ago.