I have very limited knowledge of SWMM and its capabilities and have the following inquiry. We would like to model a series of existing pumps and forcemains with varying capacities and pressure flows, all of which convey wastewater and converge into a single outlet, which empties into an existing wastewater system. There will be no stormwater associated with the model.

Our goal is to model the existing system for troubleshooting purposes to come up with a solution for certain segments that do not perform adequately.

I want to confirm whether or not SWMM would be able to model this scenario as a dry weather sanitary flow, with no rainfall input to the model, assuming that we have all the data for the existing pump curves and pipe pressure conditions, and incoming flows.

We often use EPAnet for modeling lift station and forcemain systems instead of SWMM. Just need to enter flow patterns to represent dry weather flow as a negative demand. Model wet well using a tank, etc.

SWMM is perfectly capable of handling these scenarios, and we have successfully modelled several similar cases. SWMM has a "forcemain" conduit type, and can use the darcy-weisbach equations for pressure flow calculations. You can enter all your pump types and wet well characteristics, and run a multitude of scenarios as you see fit. We ran 50-year long models on such systems with no issues. The only two things to be careful with are:

1. make sure either your nodes allow surcharging to happen (or have a really high depth) so that you are not losing your water
2. do check the stability of the flows and water levels in your pressurised system as you may need to reduce your time step to ensure smooth calculation results.

I concur with Conrad's reply.

To reassure you, here is some real-world experience...

We use SWMM (actually infoSWMM but same engine) to dynamically model complex WasteWater systems including -- multiple pumps, forced pumped and forced gravity pipes, storage vessels at wet wells, pumps auto switching, overflow & bypasses, elaborate RDII scenarios, and complex gravity systems ------ all in one model using Dynamic-wave solution methodology.

Works well - just have to be careful of math stability and CHECK the output to make sure it's working and you expect.

Mannings for unforced Grav & DW (or HW) for forced is a great feature.

My $0.02 ...

We prefer the relative simplicity of Epanet as a first cut model before jumping into the complexity of SWMM. We use Epanet to study the potential interaction between competing lift stations, look at pump run time vs. wet well sizing, investigate different pump options, etc. Very easy to look at a graph of head vs. time for a number of pumps and observe the interaction. This output translates very well to final report figures.

SWMM (either with the Manning circular pipe or Hazen-Williams forcemain option) has issues during pump startup and shutdown (transients). These exist in real life, but do complicate the analysis and graphics. Also SWMM really wants to simulate open channel flow. Try modeling a forcemain with a high point in the middle of the pipe profile. Chances are the HGL profile will break to open channel flow at the high point even though we observe closed conduit/pressurize flow in a real system. We've had the best success in hilly country by simply modeling the SWMM forcemains as a single pipe, but losing the intermediate HGL profile in the process.

Best success is using a transient model like Hammer or TransAm to model a single pump cycle, but obviously not suitable for large scale modeling or many pump cycles.

SWMM 5 will do fine with your low-pressure manifold force main. The program is routinely used to analyze strictly sanitary systems (i.e., little to no RDII), whether all gravity, all pressure, or a mix of the two. We have used InfoSWMM, whose computations are based on the SWMM 5 code, without any problems to model a large, regional, manifolded, high-pressure force main system (~80 pump stations including their gravity collectors) with pressure control facilities (think water supply booster pumps, and bleedoff valves) operated by SWMM's control logic capabilities. That model also included a low-pressure force main with 5 manifolded pump stations.

For purposes of numerical stability, I prefer oversizing pump discharge nodes a bit, favoring extra height over a larger diameter, and giving them a generous surcharge depth. Putting too much volume in a force main node can overly attenuate flows, as the volume from the pump will be pushed up into the excess storage space before the HGL differential gets the water in the pipe moving. Basically, try to keep excess system storage to a minimum while at the same time ensuring stable computations. If you start with all force main nodes (diameter and height) set to the largest connecting pipe size and a minimum time step of e.g. 5 seconds, you can increase pump discharge node size until you get stable runs, then increase the minimum time step if you want faster run times. Recheck stability, adjusting storage and time step as necessary to reach a final solution with which you're satisfied.

The number of pumps on the manifold will impact the minimum time step and, by extension, model run time. Every pump on the manifold will be impacted by every other pump turning on and off: the pressure waves being passed through the system due to pump cycling will tend to slow down the model computations. Also, the more definition you have in the (Q-dH) pump curves, the better.

The use of EPANet to analyze force mains is a holdover from SWMM 4 (and earlier) days, when SWMM was not very good for force main analysis.

For sanitary pumps stations I tend towards EPAnet based programs. Partly because of the dent still in my forehead from SWMM4 days and because of the reporting/visulization features of EPAnet are better suited for pressure systems. It also allows modelling of minor losses for bends, check valves, reducers, etc. to be a little easier. I have also found for multipump systems to be a little bit more stable (even for SWMM5). As Kirby alluded to be careful of those highpoints where a transition to gravity can occur or does not occur depending on the forcemain profile and location of air/vac valves.

I don't disagree with Kirby and Jonathan i.e.

1. Use EpaNet if just closed conduit system.
2. SWMM rising mains have problems at local high points -- but NRVs (flap valves) and initial conditions can help

BUT if you need to model entire San Sewer system you can make SWMM work -- but it does take lots of effort and checking. (for pump stations I sometimes start them at slower speed then go to full speed once WL changes significantly -- can reduce math spikes and transient effects.)

Are you saying you use STORAGE nodes for pump discharge nodes rather than junctions (which have sys default of 1.2m dia)? Or have you found a way to reduce the Dia for ordinary junctions just those on the Forced?

You can get the same depths for the pump discharge nodes using either the normal surcharged node or a storage node due to the following similarities:

1. The friction loss in the force mains and the pump curve are the dominant controlling variables for the depth at the storage node,
2. The water surface node continuity equation for the storage node (the node area is used as the denominator) is different than a surcharged node (the sum of dQ/dH for the links is used as the denominator) but a. Since the process is iterative anyway and really you are balancing the friction loss with the pump curve the stoage does not have a large effect - they are not exactly the same however, b. Because the storage node surface area is not only the area of the surface node but usually half of the area of the connecting links the storage area is usually not an important factor to tweak.

Important, friction loss and pump curve. Less important is the question of storage versus storage node - though you almost always get a higher depth with the surcharge node so it is a more conservative solution.

This has been an excellent discussion and I just wanted to add that (1) a lower time step is always a good idea for the pumps to start and as Bill states increase the time step later when you are happy with the results and (2) if you have a long rising force force main many times you get a better and faster solution by putting in a break node about 90 percent of the length of the link. The reason for this is you often have a full pipe at one end of the force main and a partially full pipe at the other end of the force main. If you put in a break node you will have a full force main for most of the length and then a 10 percent transtion to the gravity main and a partially full pipe.

We use storage nodes with a small constant and 0 coefficient.

Unless there is a compelling reason to do otherwise, we simplify the force main network to the key components (branch endpoints, multi-pipe junctions, and diameter changes).

Inserting a break node where the force main crown falls below the elevation of the force main invert at the discharge point is one way to keep the force main full. Another is to artificially sump the discharge node (if lifting to gravity) and put the force main at the lowered invert. Since the sump is below the gravity outlet invert, the force main pipe will remain full. Either method, though, requires additional labor.