Modeling Municipal Water Main Connection Using EPANET

[OaklandMechEng]

Greetings All,

I recently had to perform a hydraulic analysis of an fire main extension project for my company (we're adding 3 hydrants to an existing system). I looked around the Eng-tips forums and built a model using EPANET -- which I'd never used before.

I have data from the municipal water and power, that includes a flow vs. pressure table for the watermain we're hooking up to.

I'd like to ask:

1)how to best model this connection. So far, I have a reservoir and a pump that uses a pump curve that I built using the data I received. Not sure if this is valid or the most accurate way of doing this -- there are other types of curves and even valves I can use. I tried using a GPV valve with a head loss curve, but that gave me negative flows so I stuck with the reservoir/pump.

2) how to model a backflow preventer. Again, I have a small graph showing pressure drop vs flowrate, but not sure how to model this.

I'd appreciate any input!

 

[77JQX]

I'd make the model as simple as possible. I'd model the connection to the municipal connection as a reservoir with flow control valve, with the reservoir HGL and FCV flow at the tabular values provided, then use trial and error to find the best fit between the hydrant flow and the source HGL. I'd manually calculate the backflow loss, and subtract that from the reservoir elevation.

There are fancier ways to do this, but I suspect in the end, you're just after a yes/no answer as to whether your system will meet minimum fire flows. This should get you there.

 

[OaklandMechEng]

Thank you for the input!

However, I get errors when I connect a FCV to the reservoir -- epanet doesn't seem to like it (I get "illegal connection" errors).

In the "settings" box, I pointed the FCV to the pump curve I'd already created (I also tried changing the curve type from "pump" to a "head loss"), which also generates errors -- not sure how to implement the table of data I received from the municipality with this valve (it's just a table of pressures vs flow rates -- which looks like pump curve data, so that's why I started with the pump).

You're right -- I just want to show that we're still getting 20 psi minimum at the required flow rates (1500 gpm in my case, which I've modeled as a node with a demand of 1500 gpm and I'm taking not of the reported pressure).

Unfortunately, I don't have existing hydrant flow data to serve as "sanity check" (I've asked, the client is looking) -- ultimately, I'm not sure if modeling the municipal water main as as pump as I've done is valid or completely crazy.

Thanks again for your help

 

[coloeng]

Modeling the main supply as a pump is an acceptable method. It correctly mimics the headloss in the system as flow increases.

Good luck with your analysis.

 

[cvg]

hydrant flow data can be misleading depending on the time of day, overal flow conditions during the hydrant test, hot or cold day, booster pumps on or off during the test, changes in overall system demand since the test was done, etc. I'm not sure a hydrant test record would help much with a sanity check since there are so many unknowns.

 

[OaklandMechEng]

Thanks for the input! Luckily, the mains appear to have a lot of excess capacity so it's not a marginal case (I'd really be sweating bullets then).

Mostly, I have no prior experience with Epanet (or any hydraulic modeling software, actually) and I wasn't sure how to deal with the backflow preventers and the water main hookup. As long as the arrangement passes the "laugh test", then I think I'll be OK.

I did double check some basic parameters against charts in the Lindeburg MERM, so the numbers that I'm getting appear to be at least somewhat grounded in reality.

cvg: you're right, I hadn't really considered that. I wonder if the numbers I received from the municipality is during peak use or some sort of average or what the conditions were. In any case, I'll keep that in mind (and I'll ask next time).

Thanks again for the input forum! I'm less stressed out over this now.

 

[hydrae]

I have modeled backflow preventers using the GPV and a headloss curve extracted from the device cut sheet. Backflow preventers are a non linear headloss so the GPV is right tool.

As for the supply, a reservoir and pump with the pump curve matching the data point(s) given and using the H-W equation to get the curve shape is also how I have done that.

The negative results from using the GPV may be from not having the reservoir at the right elevation, in this method, the reservoir needs to be set at the highest posible head and then the GPV is built backwards, which can be confusing. The supply is much simpler to use a pump and pump curve with the reservoir at zero head.

Hydrae

 

[OaklandMechEng]

Thanks Hydrae,

I feel I'm getting the hang of EPANET -- from this posting and from scouring the archives.

I've actually come across (and learned a lot from) many of your posts while reading the archives and searching for "EPANET" -- thanks for sharing your expertise!

As to the valve, I did notice that I could get similar results *if* I drew it backwards, but that didn't make sense to me so I blew it away and went back to the pump. I assumed I was not doing something right.

I'll update the back flow preventer model as you suggest. I'm currently using a PBV valve that's shaving the max. pressure loss shown in the cut sheets as a conservative approximation. Luckily, this is not a marginal case so I can err on the conservative side.

Getting feedback in this thread has been a great relief.

If we were in the same office, I'd bring in doughnuts to share for everybody's help -- but I guess all I can do is wish everyone some good karma.

Thanks everybody!