Re: [Anuga-user] hydraulic structures

[Gareth D. <gar...@gm...>]

Hi Stefan,

My general advice is to be particularly careful when checking your 
results at hydraulic structures.

1) Regarding riverwalls

There is an example in 
anuga_core/validation_tests/behaviour_only/lateral_weir_hecras which 
compares ANUGA's riverwalls to a HECRAS model with lateral weirs. This 
involves flow between 3 parallel channels -- discharge is input to the 
central channel, and the neighbouring channels receive flow from the 
central channel via riverwalls. The solution in each channel downstream 
of the riverwalls is thus determined by the riverwall treatment. ANUGA 
and HECRAS agree quite well in this situation. This required a change to 
ANUGA's Qfactor described below. [Also, we had to adjust the manning 
coefficient to account for HECRAS using the hydraulic radius in the drag 
term]. But both these adjustments could be pre-computed theoretically 
from the differences in the model equations -- there was no calibration.

Regarding the riverwall parameters, yes you expect them to change 
depending on the situation. The equations are shown in documentation in 
anuga_core/structures/riverwall.py (there was a typo which I corrected 
just now). The parameter Qfactor allows scaling of the 'standard' weir 
equation, which includes a submergence correction of Villemonte (1947). 
Without submergence, we have q = Qfactor * ( 2/3 sqrt(2/3 g) H^{0.5} ) * 
(H) (which is approx = Qfactor * 1.7 * H^1.5).  The latter form of the 
weir equation is quite standard but the 1.7 constant is different 
depending on the type of structure, and Qfactor allows you to account 
for this.

For example, in the above mentioned test we set Qfactor = 1.1/1.7 
because we are comparing with a HECRAS lateral structure which has a 
weir coefficient of 1.1, not 1.7 (see their manual).

More generally, my understanding from the literature is that you may 
well expect to have to calibrate weir coefficients in realistic 
applications -- which would imply varying Qfactor.

Now, the other riverwall parameters (s1, s2, h1, h2) control how we 
transition back to the shallow water equations as the weir becomes 
significantly submerged. This is not 'standard'
in the same was as the weir coefficients are relatively standard in the 
literature. They were not changed in the test case above, but it's hard 
to offer much guidance on how/when you might adjust them. Just make sure 
you understand what is being done (from the documentation), and you can 
experiment if you expect it will matter.

Note you can also use the shallow water equations directly for riverwall 
flow (one of the beauties of discontinuous elevation), by adjusting the 
s/h parameters. In my experience this leads to similar results as the 
weir equations for non-submerged flow, but less flow in a submerged 
state than predicted by the weir equations. The latter clearly become 
unrealistic as the weir becomes very submerged, which is why we need to 
transition back to SWE at some stage.

2) Regarding culverts and bridges

The general approach is to put the bridge deck in your elevation model 
(perhaps with a riverwall if you want to treat it as a weir), and 
include flow under the bridge using some sort of operator. There are a 
number of particular culvert type operators in ANUGA. I suggest you be 
careful if using these, sometimes they are fine but on occasions I have 
seen unrealistic results.

An approach which has worked well for me is using an 
'internal_boundary_operator', which allows you to provide a function 
f(headwater, tailwater) to compute a discharge between 2 locations. This 
means you can get such internal boundary curves from theory or other 
software, and put them in ANUGA. In my application, I got the curves 
from HECRAS. There is a function for this in 
'internal_boundary_functions', along with other examples.

That said, there are still subtleties with how the momentum flux through 
such structures is treated. See a discussion here:
https://github.com/GeoscienceAustralia/anuga_core/issues/84

There are a few examples using structures in 
anuga_core/validation_tests/behaviour_only/

----

Cheers,
Gareth




On 31/03/16 23:01, Stefan Kummer wrote:
> Hi,
>
> I have some questions to hydraulic structures in ANUGA.
>
> 1) Culverts
> What is the meaning of the following arguments in the __init__-function
> of the Structure_operator class (structure_operator.py): enquiry_points,
> z1, z2, apron, enquiry_gap?
> How does ANUGA treat roadway overtopping at a roadway with a culvert?
> Is there a way to model embedded and open-bottom culverts with natural
> invert?
>
> 2) Bridges
> What is the best way to model a bridge (respectively the bridge deck)
> that may be overtopped during a simulation?
>
> 3) RiverWalls
> Are there any experiences with the default riverwall hydraulic
> parameters (see: RiverWall class  in riverwall.py)?
> - default_riverwallPar={'Qfactor':1.0,  's1': 0.9,  's2': 0.95, 'h1':
> 1.0, 'h2': 1.5}
> Are there conditions/situations those parameters should be changed
> significantly?
>
>
> Thanks,
> Stefan
>
>
> ------------------------------------------------------------------------------
> Transform Data into Opportunity.
> Accelerate data analysis in your applications with
> Intel Data Analytics Acceleration Library.
> Click to learn more.
> http://pubads.g.doubleclick.net/gampad/clk?id=278785471&iu=/4140
> _______________________________________________
> Anuga-user mailing list
> Anu...@li...
> https://lists.sourceforge.net/lists/listinfo/anuga-user