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

network dividing

Hello , 

I am modelling a large nested  chilled water network. chilled water closed loop  systems don't have demands.the pumps are used to circulate the chilled water  

I would like to know if it is right to  model the supply lines only & use demands at the location of the air handling units. that would cancels out the static head but would it exactly doubles the friction loss if the return takes the same path as the supply line. Note that my network is not straight forward as the previously showed picture. It is a nested loop as this picture shows ;

  

modelling what I really have requires taking branches from nodes , modelling the Air handling unit as a PBV for ex & making the same loop as a return loop. 

Also , is it right to divide the network into sub-networks ? how ?

Best,

  • Hello Ahmed Elshorbagy,

    I think the best way to answer your question about if it's correct to model the supply lines or not is to set up two scenarios with one modeling just the supply line and the other modeling all the lines. You can take advantage of the active topology alternative to do this, which is explained in the wiki technote found below:

    communities.bentley.com/.../active-topology-management.aspx

    In case you need some help with scenarios and alternatives you can find some information on that in this wiki technote:

    communities.bentley.com/.../4508.scenario-and-alternative-management.aspx

    After you do this you can compare the results and use your best engineering judgment given what you know about the system to determine, which setup will work best.

    For the question about demands at the air handling units you should consider if there is actually a real demand there in your system. If this is true close loop system there may not be a demand there as you have state above, which should be fine and can still be modeled in the WaterCAD software. I found this wiki technote on closed loop systems that might have some useful information for you:

    communities.bentley.com/.../modeling-a-closed-loop-system.aspx

    As far as the friction loss goes that is only calculated on a per time step basis, so I'm not sure that would double the friction loss. Also, I'm not quite sure I understand what you are speaking of in this situation because if this is closed loop system doesn't the water enter from one direction and then stay in the system and loop around. If that is the case then I'm having a hard time picturing how the return line and the supply line are the same pipe. Perhaps if you can help us understand this a little bit better we can comment on it.

    I'm not sure it's right to divide the network into subnetworks in this situation unless you think it can safely be modeled as such. If it's actually one complete network then I would take the conservative approach and model it as it is in the real world. If you did want to divide the network into subnetworks then what you could do is select the portion you want to put into it's own network and go to File > Export Submodel. Then you would name the file and create a new file. In the new file you go to the File menu > Import Submodel.

    Regards,

    Mark

    Mark

  • Hello Mark,

    Thank you for the great contribution to solve my problem.

    I will do the scenarios & post my conclusion.

    About the closed loop chilled water system , it is the same as the tech. note. the supply & return pipe are not the same. Going with the tech. note example, there is a supply line , air handling unit which is the heat exchanger & a return pipe. refer to figure below:

    the return line have the same properties as the supply & physically , both supply & return pipes takes the same route. The static head of the pump is always zero because it is a closed loop.

    I am willing to get the pump head & I have read tech. notes & other threads about how to get it. Now, my  question was ,"Is it right to model this system by replacing the PBV(that is pressure loss is zero ) with a node that has a demand equals the system flow & removing the return line (red) ? the obtained pump head would be twice the real case ?

    In this example ,it is easy to model both with scenarios & using active topology. this example also showed that "yes , the friction loss would be exactlydoubled ".

    My model is too large & have many interconnected loops. so it is hard to do both scenarios but if I was the only solution & will do it.

    Imagine that there is a municipal water network for a city & each demand reaches the end user & come back in another network similar to the supply network. In other words, You have a supply network with positive demands & you calculated  only the friction loss head. Now this demands , has been transformed to heat exchangers & I have to build the same network with same pipe sizes , elevations ..etc to return with the water from the heat exchanger to the pump station. Would the friction losses be exactly twice the first case ?

    I hope I have introduced my problem well this time.

    Best,

    Ahmed

    Best regards,

    Ahmed Elshorbagy

  • Ahmed,

    In the situation you described the friction losses would not be exactly double unless the length of pipe returning to the pump station (point B to point A) from the heat exchanger was exactly the same as the length of pipe going from the pump station to the heat exchanger (point A to point B). Does that answer your question?

    Regards,

    Mark

    Mark

    Answer Verified By: Ahmed Abdel Kader 

  • If you remove the return line and replace the PBV with a junction demand equal to the same flow previously seen, then the pump still need to pass the same flow and thus the same head. So, you should get the same pump head and friction losses.

    Here's an example case I set up, using a made-up pump curve and reservoir set to 100 feet, with a tiny 0.1 inch pipe connecting the reservoir to the pump. With this setup, the flow is about 1200 gpm and the head is about 40 feet:

    Now, if I replace the PBV with a junction that has a demand of 1200 gpm and increase the diameter of the pipe between the reservoir and pump to something large like 16 inches, I get the same pump flow of 1200 since it only needs to satisfy the demand. The pump then also adds 40 feet of head since that is the head on the pump curve corresponding to 1200 gpm. With the pipe physical properties unchanged, I get the same friction losses and thus pressure at the demand node.

    With that said, would you know what the "system flow" (the demand to place on the junction) is without modeling the whole loop? 


    Regards,

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

    Answer Verified By: Ahmed Abdel Kader