Joining sewer and drainage data for dual drainage model
I have the sewer data and drainage data, however, I am trying to figure out how to connect them to create a surface and subsurface in order to account for a more realistic flooding in the area. Can anyone direct me or assist me on how to proceed with these data?
Modeling surface and subsurface flow together is a challenge in EPA SWMM. There are several commercial versions of SWMM that have this capability. That capability comes at a price.
Modeling both has been done in EPA SWMM, but it is not a straight forward process. You need to visualize the overland flow paths when the subsurface system is surcharged to the ground level. The flow paths change as the severity/depth/magnitude of flow change. This is what makes it a challenge. Selecting the overland flow path parameters and elevations to represent a wide range of flow conditions is difficult. You then need to add hydraulic links/nodes that will represent these overland flow paths. This visualization requires you derive a channel cross section and/or storage nodes that will represent these flow paths. This is a trial and error process. Run the subsurface model and find the overflow points. Identify likely flow paths. Add the surface links/nodes and rerun. Review the results and examine the surface terrain to see how the results compare to your assumptions. Revise/add more links/nodes as necessary and then rerun. Run, review, and revise until you are satisfied with the results.
This is much easier to do in the commercial packages. Do a web search and then talk with the vendors if you are interested in pursuing this option. There may even be a free version somewhere that has been built by someone. There are also other commercial packages other than SWMM that will handle this condition.
Please see discussions here:
John's comments are very true, many more bits and pieces to worry about with EPA SWMM than with some commercial offerings that can bundle assemblies of nodes and links and hide these from the user. However, it is all possible. We use custom software developed in house to build both the roadway subcatchments, roadway gutter drainage network (transect links) and the inlet to sewer representations (Gutter Node - Outlet Link with appropriate rating curve - Storage Node - catch basin lead link connecting to sewer node), mainly because we work on larger road projects with lots of on-grade and sag inlet combinations, but entirely possible to construct these 'manually'.
The biggest unknown is the inlet rating curves (SWMM outlet links). If rating or capture curves from hydraulic testing of actual inlet grates are not available in your jurisdiction (and they don't seem to be available in most):
Play with some outlet rating curves and you will see they work the same under reverse flow (there is no negative side of the curve). This won't be an accurate representation of upwelling of a catch basin spilling onto a street but may be a rare event or close enough for your purposes.
Thanks for answering the question. I have been having a difficult time and have considered using PCSWMM. However, there has not been much progress. I have all of my sewer (inlets and pipes) and storm drains (manholes and pipes) and still dont know how they can be combined. I'm thinking of just using the storm drains (manholes and pipes) and determine my surface flow path and find away to somehow connect them to the storm drains.
Is there any suggestion on how to do this?
You always want your model as simple as you can build it and evaluate the problem adequately. Generally, I do not model all of the inlets at an intersection. I just add all of the flow to the manhole on the main sewer where the inlets connect. I then generally use the lowest inlet rim elevation as the rim elevation of the manhole to identify surcharge locations where water might leave the system. I am generally evaluating the sewer capacity in most of my studies. Usually not interested in the capacity of the inlets and whether they cause a problem. The design will later add sufficient inlets to capture the flow within the street design criteria.
It all depends on what the main purpose of your study is and how much complexity you need to add. Is it to evaluate the capacity of the main sewer and design a sewer to relieve the flooding along the sewer? Then I would not model the inlets. If it is to evaluate inlet capacity and identify where additional capacity is needed or design the inlets, then you need to add the inlets to the model.
If you also need to evaluate overland flow paths for when the sewer system surcharges, it helps to keep the underground system simple. If all the inlets are included, you need to add overland flow paths connecting every inlet and manhole along the overland flow path. You can end up with a lot of overland connections if you have 4 or more inlets at every intersection along with a manhole. You need an overland flow path between each one where flow is possible as well as any overland flow paths to the next set of inlets downstream or on another branch depending on the terrain.
I concur with John. I've also found that you need to do a few iterations and scenarios to determine just what is controlling flows into an existing system. Sometimes it could be inlet capacity that governs, other times the downstream pipe could surcharge or flood and modelling the inlets is a bit pointless (and/or makes things exceedingly unstable). This means the model for one return period may be a bit different than for another.
This is an interesting topic.
Reference: Urban Flood Mitigation and Stormwater Management - CRC publisher.
Hello thank you all for sharing your thoughts. However I am still determining my overland flow paths and I am not a sure how to accurately do it. Also, are streets improved as transects instead of lines? How can I convert them.
The purpose is that I want to do a flood map for my campus using the dual drainage and this is the first time I'm tackle such an issue. I haven't been successful on how to actual create the dual drainage model. I have my inlets, manholes, pipes, watershed and streets as GIS data. I have also resorted to PCSWMM but still having some difficulty.
Any suggestions are welcomed.
Further to John, Mike and Dr. Guo's comments...
When doing detailed 'dual drainage' modelling for high volume roadway design (major collectors and arterials), we put a lot of extra effort into modelling the roadway gutter drainage system for gutter flow encroachment into vehicle travelways, placement of on-grade inlets, etc. The actual design of the gutter and on-grade or sag inlet is done using a Rational Method approach using FHWA HEC-12/HEC-22 chapter 4 procedures in a spreadsheet. For the spreadsheet, we typically discretize the catchments into 5m lengths (measured longitudinally along the gutter). A spreadsheet is produced for each half of the roadway (i.e. eastbound and westbound) unless they are virtually identical. Note that superelevation of the roadway can mean the gutter can change sides, and this can further chop up the spreadsheet model. For simplicity, we use one worksheet to model each sag low (from one high point through a sag low to the next high point), and for each side of the road as mentioned before.
When converting these to SWMM, we typically go with 20 to 30m long catchments just to avoid having too many tiny catchments. These are paired with 20-30m long transect conduits representing half of a crowned urban cross-section roadway (extending from drainage limit on the boulevard side to centre of roadway or median). We typically place nodes with invert at gutter elevation and 1.0m depth for roadways. These transects are assumed to have vertical walls (up to 1.0m above low gutter elevation, or whatever height is appropriate) on the boulevard and centreline limits to contain flow. In practice, we try to minimize the number of roadway transect cross-sections and don't usually worry about widening the transect at turning lanes or developing too many transects to represent intermediate sections of superelevation cross-section reversal.
See the input file snippet for samples of a roadway transect in:
The SWMM roadway gutter transects lose some of the detail in local variations of longitudinal grade and crossfall, and superelevation cross-section rotation. This is OK for our purposes because the detailed design of gutter flow, flow and puddle encroachment, and inlet spacing is done within the Rational Method spreadsheet and SWMM is used for the proof of concept and investigating what happens when we couple the surface and piped (or ditch) drainage systems, how much flow spills over into adjacent unmodeled areas, etc. The above modeling includes modeling of the inlets (as OUTLET rating curves representing one or more inlet grates in either on-grade or sag placement), catch basins (as small storage nodes) and catch basin lead pipes connecting to storm sewer system. The rating curves include a design derating factor for clogging. This was described in my 2018-05-31 response post and attached links.
Points of likely overflow between adjacent sides of the road at sag low points or major overflow from the roadway to adjacent lands are modelled as several wide triangular or trapezoidal channels to link one side of the roadway to the other or spill over the boulevard. We haven't used the newer SWMM roadway overflow weir for this yet.
When conducting general dual drainage works but not specifically modelling detailed pavement drainage and inlet performance, we follow the same procedure as above except but generally increase the catchment size to 30-50m measured along the gutter, but still model each side of crowned roadways with transects. Mixed urban-rural roadway cross-sections (curbed pavement with swales in boulevards) require both sides of the roadway and swales to be modelled as separate conduits. Inlets are modelled as a simplified rating curve connecting roadway gutter to storm sewer node without the extra catch basin storage node or catch basin lead pipe. Culvert type inlets from ditches and swales are either not modelled or modelled as a short pipe.
For coarse master planning models, modelling of the roadway gutter system and small roadway segment catchments is omitted altogether. Large catchments are use with catchment width adjusted to represent length of flow path (average or harmonic mean of Subarea/Flow path length for each section of roadway draining to a sag low).
Designated major overland flow channels are modelled as trapezoidal or transect cross-sections, typically with conduit lengths of 30-50m or as appropriate. For general surface overflow, we start with a terrain model and contour plan (we typically produce this using Acad Civil 3D from topo survey or LiDAR data) and sketch the likely flow paths, convert these to nodes and conduit links (typically wide triangular cross-sections, or whatever is the best representation of the drainage system, culverts, etc.) then discretize the catchments.
Since your project is a Campus, roof drainage and rooftop detention on flat roofs or parkades will also play a part in your model. One simplistic way of dealing with this is to simply use a second hydrograph delayed by 10 minutes to simulate the routine time from roof to ground or roof to storm sewer via the building drainage systems. The non-simplistic way is a ton of work and involves modeling some of the building drainage system (usually model the horizontal building drainage to scale, but shorten the vertical parts of the system to account for plunging flow that SWMM cannot simulate). Since you have an entire campus, I'd start out with the simplistic method for modelling roof drainage and refine at a much later date if this is desired.
For modelling very large areas, we've used various tools that can take a grid-based digital elevation model (DEM) and produce a drainage map (google Tardem or Taudem). PCSWMM's 'Watershed Delineation' tool does a good job of this with minimal effort or input, but does require you to add storage nodes to correctly model impounded areas (natural depressions, or designated dry pond type impoundments intended to delay/detain outflow). If you have LiDAR data available, you can use various software tools to delete certain data elements (rooftops, medium and tall vegetation, etc.) to represent the ground surface only. We use Global Mapper software for this, and for exporting a mixed bag of contours, ground elevations, TINs into a gridded DEM. Typically need to reduce the grid interval since LiDAR point density can be way too high for your purpose - we typically use a 3m grid for large area modeling which averages out a lot of the minor jitter.
We've found the best way to organize ourselves and build models is on a cad base plan that includes XREF layers for the legal base plan, orthophotos, buildings, roadways, terrain contours, surface soil types or surface cover types, GIS info on storm sewers, and catch all 'work in progress' notes. System subareas can be converted to SWMM input format a piece at a time, or divvied up amongst modeling staff. The overall base map is a good means for you to convey progress to managers/clients. To minimize rework, I'd start with a small portion of your project area and model it completely since you will likely change your approach or understanding of a few things as you progress.
Before doing any 'real work' and to minimize risk of heading in the wrong direction, set up a few small testing models to study and test various model components that you may not be familiar with such as roadway gutter transects, inlet rating curves, roofs, etc. and verify that they function vs. first principles or hand calculations.