User Forum of Software BASEMENT

Gonne

User

Registered: 2015-11-20

Posts: 15

Very long time calculation

Hi,
I'm trying to simulate a river over a lenght of 10 km for my master thesis. I am still in steady flow simulation and I put in steady hydrograph an increasing flow rate in step as follows (because in an only time some problems occur):
0     50
1000     70
2000     90
3000     110
4000     130
5000     150
6000     170
7000     190   
30000 190

The problem occurs when the discharge reaches to the final step: the simulation slows with the increasing of the discharge and the computational time comes up to 100 hours and beyond.
How can i solve this problem?
I tried to change initial condition with:
- dry
- index table (h=1 m or h= 2 m)
I also tried to change the boundary outflow condition with:
- hq relation (slope)
- zhydrograph
but the computational time is still very long!
What condition I can change to reduce computational time?

Thanks!

Vieri

Lukas Vonwiller

Developer

Registered: 2014-09-04

Posts: 127

Re: Very long time calculation

Hi Vieri

It has to be related to your mesh.
For increasing discharge you will have additional elements that get wet.
Maybe you have some small elements that get wet for the higher discharges.
Check the 'balance' output using 'timestep' to see the element that is limiting the computational time step.

Hope that helps!
cheers, Lukas

PaoloTamagnone

User

Registered: 2016-07-01

Posts: 8

Re: Very long time calculation

Hi,

I 'm running the simulation of a river about 10 km, I created the mesh with SMS of about 152000 and 76000 element nodes .
I am using a hydrograph of about 180000 seconds .
my simulations take about 30 hours to get to the end . my PC have i5 quad-core 3.5GHz and 16gb RAM.
can i reduce simulation time in some way ? I'm already using a CFL = 1.8 .
the RTS falls rapidly and stood at values less than 2

PaoloTamagnone

User

Registered: 2016-07-01

Posts: 8

Re: Very long time calculation

Hi,

I have created a balance output and gave me this result

# time       hydraulic_timestep     Element        Edge     
0                              1                     -1             -1
3600.53241     0.681831226         148268         222343
7200.09143     0.130216644         152991         131731
10800.0804     0.111952762         153055         229783
14400.0096     0.0949327202         153055         229783
18000.0349     0.0902452229         152963         129168
21600.0594     0.0832211924         153050         229780
25200.0296     0.0719957586         153048         229777
28800.0257     0.0468099327         152956         229702
32400.0019     0.0648657682         153062         229789
36000.0419     0.0490019006         152977         229714
39600.0073     0.0627681652         153057         229785
43200.0348     0.0676396092         153055         229783
46800.009             0.0494694572         153062         229789
50400.0407     0.0575959072         153055         229783
54000.0574     0.0624672966         153062         229789
57600.0504     0.0538331087         152963         129168
61200.0307     0.057463529         152963         129168
64800.0428     0.0560265285         152963         129168
68400.0067     0.0535059395         152963         129168
72000.0034     0.0555983332         152963         129168
75600.04             0.0543317988         152963         129168
79200.0236     0.0538793388         152955         130490
82800.0234     0.0489976541         152916         229692
86400.0372     0.0455148768         152963         129168
90000.0351     0.0523729667         153048         229777
93600.0343     0.0411328902         152963         129168
97200.0104     0.0443667293         152963         129168
100800.039     0.0437701627         153049         229778
104400.011     0.0446191978         152963         129168
108000.021     0.0389414887         153048         229777
111600.023     0.0453055354         153049         229779
115200.042     0.0430634924         152963         129168
118800.024     0.0432327431         152916         229692
122400.04             0.0439261758         153055         229783
126000.009     0.0285947974         153049         229778
129600.014     0.0378440248         153049         229779
133200.004     0.0426737356         152963         129168
136800.009     0.0441424662          83211         125243
140400.012     0.0446114433          83211         125243
144000.038     0.0403724197         153048         229777
147600.006     0.0103481717          83211         125243

this hydraulic_timestep are very low.

so now I'm supposed to go to change these elements to speed up my simulation ?
is correct?

thanks.

Lukas Vonwiller

Developer

Registered: 2014-09-04

Posts: 127

Re: Very long time calculation

Hi Paolo
I assume that, with an unsteady hydrograph there will always be another element that limits your time step.
Maybe you might check your balance output using a steady hydrograph with Qmax as constant inflow.
Thereby you maybe detect an element that limits your time step during peak discharge.
However, the time steps in your recent balance output are in a reasonable range.

Maybe you can test different numbers for number_threads defined in the PARALLEL block.
Try not to use 100% of your processors (but close to) in order not to slow down your simulation.

Hope this helps, cheers
Lukas

PaoloTamagnone

User

Registered: 2016-07-01

Posts: 8

Re: Very long time calculation

thanks Lukas.
I set number_threads=0 so as to use automatically all available cores on the system.
but it is right that hydraulic_timestep is less than one second?

regards

Lukas Vonwiller

Developer

Registered: 2014-09-04

Posts: 127

Re: Very long time calculation

The time step is determined by the CFL criterion, and is in a reasonable range.
Assuming you have 8 cores, try to use different numbers of threads e.g. 6 or 7 threads.
Best regards, Lukas