Hotstart questions

[Caroline]

Hi,
I have performed a hotstart on a model. As stated, I used a flow hydrograph with fixed value upstream, and decreased the downstream water surface till my wished level. It worked well, but I am a bit currious:

When looking at the “unsteady flow time series”-plot, the most upstream positioned cross section shows a graph with a constant flow, of approx 100 m3/s. However, all other cross sections show a pattern with flow fluctuations of +- 0.1 m3/s. Is that behaviour reasonable? Shouldn’t the flow be constant all over the reach?

Appart from that, the hotstart file does not reduce the fluctuations in flow and water surface level when I use it as input (restart file) for my real modelling. Then I recieve fluctuations for over 20 minutes in the simulations before a stable value is reached. (Which is the same as the input value).
What can the reason be? Can I reduce that time?

Thanks in advance!

[Chris G.]

I think the +-0.1 meter is reasonable. While the water surface is lowering, RAS the river is continually adjusting to the change in water surface, and resulting change in storage. That shows up as a slight change in discharge. Once the water surface remains static for a while, the discharge at all cross sections should level out to an exact 100 cms.

If you are able to get to the end of your hotstart without fluctuations and instabilities, then copying the solution at the end of your hotstart run and using it as your initial conditions for your project run should also show no fluctuations and instabilities. Double check and make sure your hotstart is numerically stable-especially at the end of its simulation, and you are writing your hotstart at the END of the simulation. Also, make sure your boundary conditions (first inflow hydrograph value, initial flow, and downstream boundary) are all consistent with the last timestep of your hotstart file.

Hope this helps. Good luck.

[Caroline]

Thank you,
the hotstart now works very well as input to the simulation, eventhough the fluctuation is still there. Problem was that the boundary conditions of the real simulation.

[Author]

Posts