User Forum of Software BASEMENT

Georg Fässler

User

From: St. Gallen, Switzerland

Registered: 2018-02-06

Posts: 1

Problem with water surface elevation in flood simulation

Hello everyone

I think I first need to elaborate my problem and the history of the project I'm working on. We are currently working on flood modelling for the creation of flood hazard maps (as its common in Switzerland). In this process, we took the existing elevation model, integrated the new course of the river (with multiple culverts and bridges) and generated a valid calculation mesh with BASEmesh. The simulation worked quite nicely and the output was plausible.

HQ30 water depth
HQ30 culvert

As you can see in the previous images, a rectangular culvert (width 4.7m and height 1.8m, gate function to model upper edge of culvert) causes backwater and thereby flood retention (as intended in the project). We implemented dams (614.5 above sea level) on both sides of the channel to keep the water from flooding the nearby industrial areas. No problems so far.

We then were asked to set up an alternative model containing the existing, much smaller culvert (tube with a diameter of 800mm) instead of the rectangular one you saw in the previous picture. For that reason we changed the current mesh by hand (importing the mesh, moving the nodes, changing the elevations and placing of the new elements) to match the dimension of the existing culvert. The following simulation run (with all the same parameters, inputs and geometry, except from the culvert) showed overflowing at the culvert, which for itself could seem plausible, as the existing culvert is very small compared to the rectangular one and not capable of draining much water. The following images show the results of the second simulation with the smaller culvert implemented.

alternative HQ30 water depth
alternative HQ30 culvert

However, a closer look at the area of the culvert reveals strange behavior, especially regarding the cells where the overflowing takes place. The water surface elevation in the channel (image, point 1) and the whole flooded area upstream of the culvert ads up to approximately 613.5 above sea level, thus not exceeding the elevation of the dam (614.5 above sea level). So there should be no overflowing, right? Nevertheless overflowing occurs at the dam near the culvert. In the corner of the dam, the water depth is bigger than at nodes located at a lower level (images, point 2 and 3), exceeding the dam by 30cm on the crest while everywhere else, the water surface elevation does not reach the dam crest by far.

alternative HQ30 point 1
alternative HQ30 point 2
alternative HQ30 point 3

As an estimation, I determined the volume of water leaving the channel. If distributed on the backwater area upstream of the culvert, said volume results in an additional 0.5m head of water which leaves us at 614.0 above sea level and still below the dam (614.5 above sea level). So theoretically, we should be able to hold back the entire quantity of water which left the channel in the second simulation.

Until now, I was not able to find the reason for the specified behavior of my model. So far, I tried:
- checking the mesh for gaps or holes, remeshing the elements by hand
- simplifying the culvert to a rectangular shape of equal area
- setting a smaller value for the parameter minimum_water_depth
- using the HQ-relation instead of the gate function at the culvert
- increasing the elevation of the dam

However, the problem of inconsistent water surface elevations at the culvert remains. Do you have any ideas or assumptions as to why this happens?

best regards,
Georg

Last edited by Georg Fässler (2018-10-18 14:47:25)

Stephan Kammerer

Developer

Registered: 2015-03-31

Posts: 67

Re: Problem with water surface elevation in flood simulation

Since your model geometry is very complex, its hard to tell from distance what migth be the precise reason for the observed behavoir, but I have some generell comments on your "valid compotational mesh" which might also be an explanation:

To me it seems that you tried to put to much detail into this mesh. Its not wise to resolve every building outline in full detail (for example) . In modelling you need to abstract reality, which means that you should simplify buildings thereby avoiding very small elements and a generally irregular mesh. Does this mesh fullfill quality criteria? Did you check "Conforming Delaunay" in BASEmesh and use a reasonable "Minimum trangle angle"? To me it seem the mesh contains very acute angles.

Generally BASEMENT is a robust tool and your simulation also runs on "questionable" meshes, but I'm not too astonished, that some unexpected behavoir might accur in such situations.