Pipe model calculates unreasonable pressure drop when different diameters are used

[dietmarw]

A user reported that when one sets a different ouput diameter then the input diameter of OpenHPL.Waterway.Pipe then one can yiels pressure drops which are larger than the actually headdifference which is impossible.

[dietmarw]

@boerrebj This might be of interest to you. I've tried to improve the handling with ef0b3ac, though it still generates abnormally high pressure drops when a pipe's diameter is reduced. Any input is more than welcome. It would be nice to be able t have a simplified pipe model that is not so calculation intensive but still be able to have it tapered.

[pandeysudan1]

A user reported that when one sets a different ouput diameter then the input diameter of OpenHPL.Waterway.Pipe then one can yiels pressure drops which are larger than the actually headdifference which is impossible.

I have explored this a bit (could you please share the parameters of the system so that we can reproduce it as a Minimal Working Example?).
Here are my observations and assumptions.

• It is completely fine to have the outlet pressure greater than the inlet pressure. This results in a negative Δp, while the velocity remains positive.
  You can observe this by experimenting with the intake pipe in Examples.Simple.

• I simulated the intake of Simple.mo for various height differences (H): 25 m, 15 m, 5 m, 2 m, 0.5 m, and negative values.
  The outlet pressure becomes greater than the inlet pressure when the head loss (due to friction and expansion/contraction) exceeds the static head difference.
  The cutoff occurs around a 2 m head difference.
  (Currently, there is no option to plot the head loss directly, but we can plot the frictional force, which unfortunately does not clearly visualize the head loss.)

• It just occurred to me that point 2 likely describes exactly the case being discussed in this issue.
  If it is something different, please create a Minimal Working Example (MWE) so we can investigate further.

• While developing the Pipe model—which is intended for subsystems such as the intake, headrace tunnel, penstock, and conical draft tube—we considered it as a circular pipe with diameter D.
  The model works well in this form without including F_taper.
  For other pipe geometries such as dome-shaped headrace tunnels or rectangular closed flows, the hydraulic diameter (Dh) should be used instead of the mean diameter D, as is standard practice.
  For diffusers, nozzles, or other tapered pipes, additional friction losses due to tapering should be included.
  For these specialized or irregular geometries, we may need to introduce correction factors into the model.
  I had previously implemented such corrections in the surge tank (other than SimpleSurgeTank) and draft tube models.

Please correct me if I have misunderstood the core idea, @diermarw.

[dietmarw]

@boerrebj already started working on a solution. So it may well be that this issue can soon be closed with a more robust pipe model.