I am modeling a downtown, urbanized area with long culverts in a 2D environment. I am using a steady flow hydrograph to represent the 100-year event. There are several split flows that occur for the 100-year event making a 2D environment preferable. The area was originally modeled in a 1D environment using bridges to represent the long culverts. I believe this was partially done because the culverts are open-bottomed and have irregular-shaped bottom sections. Whether or not using bridges over culverts is appropriate is up for debate. Replacing the bridges with culverts in the 1D model seems to generally produce lower water surface elevations.
Now for the wormhole part - I am using wormhole culverts to represent the culverts in the 2D model. Some of the culverts are quite long (>200 ft), and I wanted to share my observations. When I chunk out the 1D model to represent the culverts separately, assign a tailwater elevation and flow from the 2D model, and then compare the 1D headwater result to the 2D headwater result, I notice that the 2D model is producing higher headwater results. The 2D headwater is higher by 0.14 ft to 0.54 ft.
I should mention that I am having some issues with getting a perfectly stable simulation. Some culverts see flow fluctuations of 5 to 15 cfs for a total flow of 300 cfs. Could this potentially be the cause of the differences between the two models? Additionally, I have given some thought to the applicability of using the same Manning's n values between a 1D model and a 2D model. I have not seen a lot of discussion regarding wormhole culverts and would be curious to see what others have found and if they have any explanations for their results.
Re: Wormhole Culverts - Discussion and Observations
Thanks for sharing that Jennifer. It's common (and likely) to see differences from 1D to 2D, if for nothing else, because the 2D equations account for more of the losses that we tend to wrap into the n value for 1D. Assuming both 1D and 2D models are stable, I would tend to believe the 2D results more. But there are other things to consider as well. Cell size can play an important part in this. If your cells are too large across a conveyance path, this can cause a smoothing of the velocity profile, resulting in lower discrete velocities and higher water surface elevations. I generally target 5 to 7 cells (minimum) across an important conveyance path (e.g. across a main channel, or even just a concentration of flow in an overbank area). Of course, stability is the first thing to work out. Make sure you are targeting a Courant number of around 1 as much as possible in your 2D area. This will usually take care of most of the stability issues, including the pulsing of flow you mentioned.
I'll throw this out for consideration. I have a model that I'm modeling the channel in 1D and have it tied to a left and right overbank 2D area. Along the length of the channel are some long culverts. What I did was stitched the 2D area together over the culvert so that flow could leave the channel and flow in the 2D area and be able to spill between the 2D areas along the culvert. This was an existing model that I added 2D to so the culvert deck was already coded in.
When I made my first run I realized that I was getting a "double counting" of flow leaving the channel due to the 2D area taking flow and the 1D also accounting for weir flow over the culvert deck. I went back and raised the deck to an artificially high elevation so I could eliminate 1D accounting for weir flow the profile was closer to what the 1D only was showing for a headwater.