MRN: tracer output

MOHID Land forum. Questions and discussion related to MOHID Land
davidbrito
MOHID Advanced user
Posts: 187
Joined: Thu Jan 01, 1970 00:00

Re: MRN: tracer output

Postby davidbrito » Tue Apr 23, 2013 15:26

I have done some math and your dicharge only lasts for 6 seconds.
And in 6 seconds the discharge is only 40kg.

You should do a discharge that lasts at least some time so that the duration of the timestep is not relevant.
For instance if the timestep at that point was 200s than the concentration value and flow will be used for 260s and the mass will be 1800kg.

So I suggest you to do discharges of one hour or even more so that the timestep will not have influence and the discharge time is closer to the input.
In 3 hours (10 800s), dt's of 200 seconds make no much difference. Of course you can also limit the max timestep to 1s or so but your simulation will take loooooong to run.

Best regards,
David

epelde.ane@gmail.com
MOHID Advanced user
Posts: 172
Joined: Thu Jan 01, 1970 00:00

Re: MRN: tracer output

Postby epelde.ane@gmail.com » Wed Apr 24, 2013 10:31

The discharge lasts 0.1h, so 360s, no? although it was 6s, being the mass flux 500g/s, how have you obtained 40kg?

And, I don’t understand well what you say about time step and discharge duration. You mean that the model doesn’t consider nothing happening if its duration is shorter than time step? Then, you say “For instance if the timestep at that point was 200s than the concentration value and flow will be used for 260s and the mass will be 1800kg” I don’t understand why you say that time step is 200 and then is used 260.

Integrating mass and flow, I got 1800kg, so, if the duration of the discharge was longer than 360 I would recover more mass?
With the discharge I´m considering (posted on 15 april) I´m introducing 2480 kg, right?

And, the tile of the concentration line can be explained without diffusion?
Thanks a lot,

davidbrito
MOHID Advanced user
Posts: 187
Joined: Thu Jan 01, 1970 00:00

Re: MRN: tracer output

Postby davidbrito » Mon Apr 29, 2013 18:30

Hello,

I may have seen wrong bu if the discharge lasts 0.001h than it is 3.6s.

If the time step is the order of 200 or 260s at that period and it gets the discharge value it will use the flow and the concentration during that dt. And the amount exported will be around 1800kg. You have to think that the things that model reads are maintained during dt because the model only updates things from dt to dt, during the dt the things that are read (like flow and concentration) are maintained during dt. That does not pose any problem in river discharge or WWTP discharge where the discharge frequency of variation (days or hours) is muuuuch longer than model dts (some seconds).
But when you discharge something in 6 seconds (or even in 360s) than the dt may play a role to "change" the input load.


If you want to do math please do a long discharge.
You can do the same math if your discharge lasts for example during 3 or 4 hours or more why not a day? and not seconds that is the same frquency as the dt and the error increases a lot.


We will talk about concentration values after you do a longer discharge because they can be affected by this super fast discharge. Use the last .exe that we have sent you.

Best regards,
David

epelde.ane@gmail.com
MOHID Advanced user
Posts: 172
Joined: Thu Jan 01, 1970 00:00

Re: MRN: tracer output

Postby epelde.ane@gmail.com » Thu May 02, 2013 10:34

Hello david!

Ok, so I understand that the model reads the discharges after the time discretization and so, the discharge has to last at least the dt. So related to this, I wonder if I do the following thing right:
To see more or less the time step, I open any output file and I see the seconds that last the intervals at which I´m introducing the discharge, right? Which in this case, are of about 4000s.
But, then, in the output log, in last output, I see that max time step so far is 191s.
Can you please tell me the relationship of these two numers? or why the max time step at the end of simulation is smaller than a time step inside a data file?

And just another doubt. I´m introducing a pulse which I´d like to be considered as a step, without interpolation. I´m using USE_ORIGINAL_VALUES: 1 inside the block of the discharge, but then, I see in the graph of the integrated mass that the model is interpolating data. I´m using that keyword in the right place?


thanks again!

davidbrito
MOHID Advanced user
Posts: 187
Joined: Thu Jan 01, 1970 00:00

Re: MRN: tracer output

Postby davidbrito » Thu May 02, 2013 18:26

Hello,

One thing is time step, the model time step, that in case of MOHID Land is variable.
This is the time between computations it goes from a few seconds to hundreds of seconds.

Another thing is the frequency of output. If you are looking to output files (HDF or timeserie) than you are looking to instantaneous results and they do not get all model dts because model dts are far smaller. Output frequencies are usually hourly or daily so they are much less frequent than the model computation intervals (model time step).
In outputs you are omly seeing model results at a constant frequency that is much bigger that the frequency that the model computes things. The maxdt in output window is the max interval between computations.

The frequency in results you choose in OUTPUT_TIME (inside Modules data files) for HDF's and in DT_OUTPUT_TIME for timeseries (inside the timeserie file).


The keyword USE_ORIGINAL_VALUES : 1 has to be inside the diacharge block so it seems right.
I hope that you are using a discharge as a pulse but now for a longer period.
What makes you say that the concentrations is interpolated? Remeber that concentration is not interpolated from the time serie (it is used a ladder approach) but that does not mean that mass will be ladder shape because you have flow in the river that can be changing. Only concentration changes in ladder way, not forcingly mass, that is very important.

Best regards,
David

epelde.ane@gmail.com
MOHID Advanced user
Posts: 172
Joined: Thu Jan 01, 1970 00:00

Re: MRN: tracer output

Postby epelde.ane@gmail.com » Fri May 03, 2013 08:24

ok david. undestood the dt problem! :)

Now, the discharges are read in the nodes, while outputs in river are output for reaches, right? So the concentration (with corresponding given Q) I give as a discharge for a time step, is going to be updated for the Q is flowing in the reach for that time step, right?

So, althought I input a constant concentration discharge, as output nitrate concentration is affected by the water in the river can be of different concentration, no? but the mass should show a perfect step, no? wherever the reaches are. Cause if there is no diffusion, all mass should flow with water velocity and should reach the reaches at different moment but with same value, no? Even in downstream nodes, no?

I post here some images so that you can have a look.
Attachments
mass concentration.png
mass concentration.png (48.81 KiB) Viewed 6428 times

davidbrito
MOHID Advanced user
Posts: 187
Joined: Thu Jan 01, 1970 00:00

Re: MRN: tracer output

Postby davidbrito » Fri May 03, 2013 09:51

The concentration is evaluated in nodes (just as water level, water depth, etc) and discharge is evaluated in the reach.

When you discharge something it will mix with the water already in the river (in the node) and then it is transported in between nodes with flow.
You see that the concentration slight changes probably because the flow in the river is not completely constant or you have a mix with water with different concentrations (in same places water can even go upstreams and do that mixing). And you have not complete constant concentrations so mass can arrive also with some fluctuations. But the fluctuations are minimum (it seems like 1%).

If you want to see perfect curves you have to test this in a constant slope channel and the river with a constant flow (discharge) so that the effects of changing flow or even backward flow or flow exiting the section and entering at later time is eliminated.

Hope it helpes.

Best regards,
David

epelde.ane@gmail.com
MOHID Advanced user
Posts: 172
Joined: Thu Jan 01, 1970 00:00

Re: MRN: tracer output

Postby epelde.ane@gmail.com » Fri May 03, 2013 11:16

thanks, I think I see...let correct me please:

when the discharge begins, the mass aported to the reach is mix with the water in the reach so that it has a "X" concentration which value goes increasing as the discharge goes on. When discharge finishes, the water of Y concentration moves down stream by advection but I see a tile in concentration graph cause the output is a kind of average for the reach and so, I see that tile, and not due to dispersion o diffusion (as I have them deactivated). Is this right?

In that case, I´m not going to be able to simulate a contaminant (which doesnt react with anything and it is like got carried away) crossing rapidly the streams, right?

davidbrito
MOHID Advanced user
Posts: 187
Joined: Thu Jan 01, 1970 00:00

Re: MRN: tracer output

Postby davidbrito » Fri May 03, 2013 13:09

I am not sure if I understand what is a "tile".

1) if a tile is a "smooth" reduction of concentration instead of an abrupt change to zero:

When the dicharge stops than the concentration has no source but is being transported downstream. after discharge end, since the mass is not transported to the neighbour node all in one time step (that would be the case of Courant 1) you do not see a instant concentration going to zero. Since in one time step, less than all mass goes to the next node, then the remaining mass is ditributed by all the volume of the node (the concentrations is always homogeneous in the node) and the concentration reduces. This is called "numerical diffusion" that comes just because we are talking about volumes and they are homogeneus. And even that you do not have diffusion if Courant number is not 1 everywhere than you have that effect. And is not very important in your case.
I tought that you were talking about these that the curves were not squares and that the maximum value during discharge was not totaly constant (had fluctuations) that is why my answer was to test this in a controlled environment and you will see that the curves mantain the maximum value and if courant is 1 they would be squares (the arriving and leaving of concetration would occur almost instantaneously).


2) if a tile is the concentration reduction to downstream

as you can see the mass does not change from upstream to downstream. Concentration reduces because volume increases. This is the effect of dilution.
As you go downstream more and more tributaries arrive with clean water so the mass is diluted in higher volumes and you get lower concentration.


MOHID Land is totally suited to simulate that transport of tracers. I hope to have asnwered all because to me those results seem normal. Mass is maintaned, concentrations reduces as volume increases downstream. The complete "squareness" of the curves you only get in stable conditions.

Best regards,
David

epelde.ane@gmail.com
MOHID Advanced user
Posts: 172
Joined: Thu Jan 01, 1970 00:00

Re: MRN: tracer output

Postby epelde.ane@gmail.com » Fri May 03, 2013 13:46

Ok david, everything clear! thank you!! :)

about "tile".. sorry I wanted to say "tail" lol..

Regards,

davidbrito
MOHID Advanced user
Posts: 187
Joined: Thu Jan 01, 1970 00:00

Re: MRN: tracer output

Postby davidbrito » Fri May 03, 2013 14:05

That is OK I was not getting the "tile"

Good to know that is clearer.

Best regards,
David

epelde.ane@gmail.com
MOHID Advanced user
Posts: 172
Joined: Thu Jan 01, 1970 00:00

Re: MRN: tracer output

Postby epelde.ane@gmail.com » Fri May 03, 2013 14:46

another one :)

Taking into account that nodes are sections and reaches are volumes, just to have this clear, I´d like to know the following points:

-Previously you said that concentration, water depth and everything is outputted at nodes. How does the model obtain that value? Considering the previous and following reaches? And then estracting the data? Cause if it happens the dilution it will happen in reaches, no? So, why in the channel flow file the columns correspond to reaches and not nodes?

-And, you said that discharge happens on reaches, but when we create the discharge file, we have to choose nodes and not reaches, why?

thanks,

davidbrito
MOHID Advanced user
Posts: 187
Joined: Thu Jan 01, 1970 00:00

Re: MRN: tracer output

Postby davidbrito » Fri May 03, 2013 17:04

Nodes have the section info and is in node that volume is accounted, not in reaches.

Think the node as the cell in the model. The reach is between nodes and is where flow is computed.
The properties (height, volume, concentrations) all exist in node and are tansported with flow (computed at reaches). Reaches only exist for computing flow, forget about properties existing in reaches and your doubts will disappear.

Discharge is not the same as river flow. Discharge is done to a node or from a node and changes node volume. River flow is computed in reaches between two adjacent nodes.

Channel flow is in reaches because is the only thing that is computed in reaches. So output file gives you a reach and not a node.
Everything that is in node appear in timeseries with node.

Hope that it is.
Best regards,
David

epelde.ane@gmail.com
MOHID Advanced user
Posts: 172
Joined: Thu Jan 01, 1970 00:00

Re: MRN: tracer output

Postby epelde.ane@gmail.com » Mon May 06, 2013 10:08

Ok :) Thank you a lot!


Return to “MOHID Land”

Who is online

Users browsing this forum: No registered users and 1 guest