Jump to content
TUFLOW Forum
Robin.Green

Circulating Flow between ISIS & Tuflow

Recommended Posts

Hi there

I am currently working on applying a Tuflow 2D domain to an existing 1D ISIS model. I have removed the 1D representation of the floodplain and successfully got the model running . However, past a certain point, the Qi, Qo and dV fluctuate dramatically suggesting that flows are circulating between the 1D and 2D domains - the ISIS long profile also shows a 'saw tooth' water profile. A review of the results shows that the 2D flows/velocity patterns are uniform up to the point, but then go 'bananas' (for want of a better word) - I have attached a 'before' and' after' image. The unexpected flow patterns appear to originate from a downstream a railway embankment before emanating through the rest of the model. The ISIS model runs smoothly as a stand alone so I do not think this is the problem.

ISIS does not report any poor convergence issues and Tuflow does not report any negative depths, or instabilities, and the CE% is within +/- 1%.

Following the advise in the manual I have followed the steps summarised below:

-increased the resolution of 1D nodes at the embankment

-decreased the timestep from 1/2 cell size to 1/4

-moved HX boundaries away from the embankment

-smoothed the topography at the embankment

-tested alternative cell sizes; base model 5m grid tested 4m and 6m.

The only thing which has had a noticeable impact was reducing the timestep, but even reducing this down to 1/4 proved insufficient. Can anyone advise if there is something else I could try, or whether it is appropriate to reduce the timesteps even further? I’d link to avoid this as the runs are long enough as they are.

Thanks in advance for any advise

post-2699-1285338333_thumb.jpg

post-2699-1285338351_thumb.jpg

Share this post


Link to post
Share on other sites

Hi Rob,

Well, you've certainly been going through the textbook with trying to fix this issue, well done! Fortunately, there's one more thing I can think of, which looks like it might apply to your model.

TUFLOW is in charge of the flow exchange over an HX boundary, but knows very little about the geometry of the 1D elements. In particular, it does not know the width of the ISIS channel, so has to make assumptions about the volume of water/storage available there. In particular, it assumes there is at least one 2D cell width of water/storage to draw from/flow into. Indeed, if you've got HX boundaries on both sides, then because there are two boundaries making this assumption, the 1D channel is required to have volume for 2x cell widths! It can be, if there is not a significant amount of flow exchanged over a boundary, that this will not cause a problem. But it can be that having too narrow a channel linked to the 2D will cause exactly the problems you are seeing.

So in your case, immediately downstream of the embankment, I assume a large volume of water is coming through the opening then spreading out onto the floodplain, thus leaving the 1D to enter the 2D. TUFLOW receives the water level from ISIS, and calculates the flow to be transferred over the HX line, but makes that assumption of 2x cell width of volume available to draw from in the 1D. If that volume is not present in the 1D, then when ISIS removes that flow (at TUFLOW's instruction) the water level will drop too far, and may drop /below/ the water level in the 2D. The next time-step then, TUFLOW gets the new low water level from ISIS and calculates a volume of water to send back, assuming there's the volume of storage to receive it;if that volume is not there then the ISIS level will rise too far, perhaps to be above the 2D water level again. And so it oscillates.

The fix then, is to broaden your ISIS channel above bank top until it is at least 2 cells wide. You need not do it for the whole model necessarily, but along the length where the problem occurs, so in this case perhaps for the whole length shown in your images. This probably needs to be in conjunction with your other fixes. If the channel is already more than 2 cells wide then please try the below!

For the same reason (lack of storage relative to that assumed by the boundary), it can be that adding additional nodes to your 1D immediately downstream of your channel is making things worse. If you've a short reach of say, 10m, but it's linked to three of your 5m cells on each bank due to where the HX lines have fallen, then this same problem can arise. It doesn't happen often, but it might be that carefully choosing your reach lengths and boundary links will make all the difference here (so that your 1D elements are multiples of 5m long and are linked to the matching number of 2D cell); for most models you won't have to worry to that measure of detail!

I hope the above makes sense and is of use to you. Let us know how you get on or if there are any other issues.

PHA.

Share this post


Link to post
Share on other sites

Thanks for the speedy reply, and the explanation.

Just for clarification - the problem occurs upstream of the embankment where water is pooling behind the structure (sorry for the poor explanation), but I guess that your explanation and solution still apply; as the confined conditions and large flows may exacerbate the situation further?

The 1D channel width is roughly 2x cell size, but due to the large flows and relatively deep flood depths (1D channel depth= ~1m, 2D flood depths= ~1m) perhaps it would be worth extending the ISIS width further to provide some more volume?

I also note the need to double check reach lengths against HX boundary lengths - I think I'm going to have a busy Monday morning!

Many thanks for the help - I'll let you know how it goes.

Cheers

Rob

Share this post


Link to post
Share on other sites

Just an update on this post - the HX liness were compared with the 1D channel length, there were only marginal differences, but these were corrected. The 1D channel widths were also checked - these were largely just over 10m (2 cells), but adding an additional few meters to the channel width was shown to resolve the circulating flow.

Thanks again for your help!

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...