This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

SewerGems - Pressure pipe friction loss very wrong

I'm working with an odd pipe configuration (an inverted siphon which discharges into a basin, controlled by an overflow weir).

Elsewhere in my model, I have some channels that I'd like to calculate at a high resolution (0.2 feet, ideally, but I can coarsen it some).

The inverted siphon friction losses are significantly incorrect and the error seems inversely linearly proportional to the calculation distance. The system is something like this (and I have attached an example):

[Manhole with loads] ---------[Pressure pipe] ---------------[Outfall with stage-discharge relationship]

To reproduce the error, change the calculation distance. On my machine, as I drop the calculation distance from the default 50' to my target (0.2') I see the headloss in my pipe rise from 0.8 ft (should be 0.5 from hand calcs) to about 8.00 ft. 

What gives? It seems maybe additional headloss is being artifically generated in each 'calculation section' of the pressure pipe - how do I get the solver to not do that? 

Best,

Sam

Error In This Pipe.stswpkg.zip

  • Hello Samuel,

    Usually an inverted siphon is modeled with conduits, not pressure pipes, Pressure pipes are typically modeled downstream of a pump and will discharge into either an outfall (if it is at the end of the system) or a manhole (if it discharges into a gravity subnetwork). I would recommend using a conduit when modeling an inverted siphon. Conduits can be modeled as being under pressure. This link has more information on modeling these: Modeling an Inverted Siphon.

    Do you have information on the hand calculations you are doing? Are the hand calcs based on this small model or based on the actual model?

    Regards,

    Scott

  • I'll take a look at that link but to answer your question re: hand calcs - the model I attached has the correct flowrates, diameters, and boundary condition.

    The "hand calcs" I'm referring to is just using either the hazen-williams equation or the full-pipe manning's equation to determine the headloss in the pipe in that model. That small model I sent is derived from the full model - I just deleted everything upstream of it - and the 'headloss' is the same across that pipe in both the full version and this 'mini' version. 

    Specific details on the hand calc.s would be...

    C=100 or n=0.013 (for hazen or mannings, respectively)

    D=42" 

    Q=13.39 MGD

    Length = 80 ft

    From mannings (V in ft/s, R in ft, L in ft):

    HL= (V*n / (1.49*R^0.67))^2 * L

    HL= (2.15*0.013 / (1.49*0.875^0.67))^2* 80

    HL~= 0.03 ft

    From Hazen  (Q in cfs, D in ft, L in ft):

    HL = (Q / (0.432 * C * D^2.63))^1.85 * L

    HL = (20.71 / (0.432 * 100 * 3.5^2.63)) ^ 1.85 * 80

    HL ~= 0.05 ft

    Then there's a minor loss coefficient of 0.25 (a 90 degree pipe bend), which adds an additional:

    HL_m = K * V^2 / 2G

    HL_m = 0.25 * (2.15^2 / 64.4)

    HL_m ~= 0.02

    So depending on friction method, I should be seeing between 0.05 and 0.07 ft of headloss (0.06 is the upper bound actually, since both the H_m and the hazen losses are rounded up in the above). 

  • Ok, I've changed the link to an inverted siphon. The problem is now approximately 267% worse - headloss is now calculated at 0.14' (compared to 0.08' with a single pressure pipe and 0.05' in 'reality')

    Error In These Two Conduits.stswpkg.stswpkg.zip

  • Hello Samuel,

    I am seeing headlosses in the range of 0.08 and 0.06 ft in the model with the conduits. Can you confirm if that is the most up-to-date version of the model?

    Another item of note would the the values involved. The model values of 0.08 feet and the hand calcs of 0.05 feet are different by about one-third of an inch. Factoring in things like rounding, accuracy of the data inputted into the model, and modeling noise, the results do not seem to be that far apart. Given all of that, the model results compared to hand calcs may be well within a margin of error.

    Regards,

    Scott

  • Additionally, you may get better results if you use a pressure junction instead of a manhole (and in general, use a pressure junction with pumps/wetwells, and conduits for gravity-flow, even if it becomes surcharged).

    You may also see improved results with the Explicit (SWMM) solver (though you will need to change the outfall boundary type) in case the implicit solver is not working well in complex situations.


    Regards,

    Jesse Dringoli
    Technical Support Manager, OpenFlows
    Bentley Communities Site Administrator
    Bentley Systems, Inc.