Jump to content
TUFLOW Forum
Tina O'Connell

Replicating MIKE 21 results

Recommended Posts

Hello,

I have a purely 2D MIKE 21 HD model for which I am trying to replicate resulting levels for a range of flood probabilities. It has an upstream QT boundary, downstream HT boundary, and starts from a constant IWL on a regular 5m grid centring on two existing bridge crossings. The ZCs were created from the MIKE 21 bathymetry file. The ZUVH points were interpolated. The Materials file was produced from the MIKE 21 resistance grid.

There is an area downstream of the bridges where the TUFLOW model consistently (for viscosity formulations constant, smagorinsky and a combination of both, for a range of viscosities from 0.2 to 1) shows lower levels in the 100 year event (I haven't progressed to smaller events so can't comment there) than the MIKE 21 model. Froude numbers in this area are approaching 1.

Could someone give some discussion on which TUFLOW commands I should be considering changing to more closely replicate MIKE 21 levels. Here are some of those I am considering /have considered:

Cell Wet/Dry Depth == 0.02 MIKE21 has drying depth 0.02/flooding depth 0.03, levels downstream of bridges are insensitive to changes in this parameters from 0.02 to 0.03

BC Zero Flow == OFF

First Sweep Direction == POSITIVE

Bed Resistance Cell Sides == INTERROGATE Switching to Average M/n makes the TUFLOW results in this area even lower

Wetting and Drying == ON NO SIDE CHECKS

Supercritical == ON Switching to OFF made minimal difference downstream, dropped levels slightly upstream (2mm) due to changes at bridge abutments

Froude Check == 1.

Free Overfall == ON

Global Weir Factor == 1.

Shallow Depth Weir Factor Multiplier == 1.

Shallow Depth Weir Factor Cut Off Depth (m) == 0.0001

!Latitude == -27.41 In MIKE 21 files, Coriolis forcing is not checked, TUFLOW results not sensitive to switching it on

Water Level Checks == ON

Oblique Boundary Method == ON

Boundary Treatment == METHOD A

Line Cell Selection == METHOD D

Inside Region == METHOD B

Density of Air == 1.25

Density of Water == 1025.

Wind/Wave Shallow Depths == 0.2, 1.

Thanks

Tina@dks

Share this post


Link to post
Share on other sites

Hi Tina,

It sounds from your description like your issue is a fairly localised one (limited to the area downstream of the bridges), and many of the broader model settings that you mention are unlikely to have much impact. Given the insensitivity of your model to eddy viscosity changes, I would guess that the best place to start is with the modelling schematisation of the bridges. If you are modelling the bridges in 2D using FC attributes, have you tested the effect of additional form losses, side wall friction, and cell flow width? Or are the bridges modelled some other way?

I don't have a working knowledge of Mike21, but my understanding is that bridge pier losses are applied using a different method to TUFLOW. It would probably be worthwhile cross-checking the head loss through the bridge structures against Hydraulics of Bridge Waterways and/or other software such as HEC-RAS (see section 4.7.1 of the manual) to help decide whether the flood levels produced by TUFLOW downstream of the bridge are reasonable.

Rhys

Share this post


Link to post
Share on other sites

Rhys,

Thanks for your reply. I have bridge decks way above the 100 year flood levels and a grid size small in comparison to the overall bridge length (5m for a 180m bridge) so wall and roof resistance are not coming into play and cell width is not reduced. All the FCs are attempting to model is the additional form loss due to pier resistance. MIKE 21 does indeed apply pier resistance differently from TUFLOW in that you specify pier size, shape, invert and obvert at a specific location and the program internally calculates the turbulence losses. As far as I can determine, it is impossible to plot the magnitude of these losses or the coefficient of loss calculated within the MIKEZero or MIKEToolBox framework, so the only reference to whether TUFLOW is operating similarly through the structures is in replicating current speeds and water levels. Having said that, it is difficult to replicate these when the downstream tailwater level is not consistent. I will persevere in changing eddy viscosity and its application to raise the downstream tailwater levels.

Regards

Tina

Share this post


Link to post
Share on other sites

Hi Tina

How different are the water levels in the area downstream of the bridge?

If the Froude numbers are approaching one, the two schemes may well differ in how they treat any transitioning to supercritical flow, so this may have an influence.

The turbulence (viscosity) representation may also be the reason. There were certainly issues with MIKE21's representation of this term in the 1990s when we were testing it, but these may well be resolved now.

We've converted a number of MIKE21 models and generally get very good agreement with most levels with a several cm (the main exception has been around structures, which seems to be due to the limited options available to modellers for representing structures in 2D in MIKE21). Older MIKE21 models (not sure whether there is now other ways) seem to use an embedded "weir" equation to restrict the flow between 2D cells at structures, and/or use a higher roughness, so you may want to cross-check whether this is the case in your model.

Good luck!

Cheers

Bill

Share this post


Link to post
Share on other sites

HMax in the 100 year event is 30mm lower.

They have only pier resistance in the model. No weir flow or higher roughness at the relevant 2D cells.

What would you decipher "Constant 0.2 m^2/s Eddy Viscosity - Velocity based" (as taken from the attached print screen of the MIKE 21 simulation file) to mean? I've included extracts from the HD manual as an attachment. The more I think about it, the "Velocity based" part would only be for Smagorinsky Formulation, the pre-processor is just not fine-tuned to "grey it out" for Constant Formulation.

post-2084-1226693455_thumb.jpg

Eddy_Viscosity_in_MIKE_21.doc

Share this post


Link to post
Share on other sites

Hi Tina

30mm difference could easily be accounted for by a different representation of viscosity (or different representation of other terms), especially around a structure where velocity gradients would be greatest. Also, if 30mm is the most significant difference I would regard this as a good match (you'll most likely see greater differences around structures when comparing with, say, a finite element model, and much greater differences if trying to compare with a 1D model!). 30mm is also likely to be insignificant compared with the uncertainties associated with the accuracy of the hydrology, bathymetry, roughness values, etc.

Probably best to approach DHI on the meaning of "Velocity based".

Cheers

Bill

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...