5.1.009 bug: missing GW recharge from LID infiltration

David Sample

We (VT urban stormwater group) have been working with EPA SWMM and PCSWMM the past few months, using LID modules, continuous simulation, and groundwater. My graduate student, Rachael Johnson (SWMMlistserve membership still pending) has compared the latest versions of PCSWMM and EPASWMM and found what might be a discrepancy in EPASWMM. Can someone shed some light on why there is a difference? What appears to me is that the infiltrated water from the LID practice is effectively "lost" in EPASWMM as it stands, and is not added via recharge to groundwater, whereas it is in PCSWMM.


Rachael's post:

When I first wrote this post, I was running EPA SWMM 5.1.008. I tested the problem with 5.1.009 to see if it was resolved. The behavior of the graphs is slightly different, but there is still a very clear problem with the LID controls and the groundwater routine.

I was trying to investigate the effect of water table proximity on the infiltration/exfiltration capacity of BMPs created using the LID controls. I was using a bioretention cell in particular. My model is very small, just one 100% impervious subcatchment routed to a subcatchment occupied 100% by the bioretention cell. The bioretention cell sends overflow to one drain and underdrain flow to a different drain, in case that matters.

I cannibalized a presentation I made concerning this topic and provided images and descriptions in an Imgur album located here: http://imgur.com/a/72yFd Again, a reminder that those graphs were generated with 5.1.008 and there is slightly different behavior now, but the overall problem persists.

For those without access to Imgur, I will explain as best I can with words.

I set my model up and everything worked as expected without groundwater. I added groundwater in and set placeholder coefficients/exponents in the lateral flow area (A1=0.1, B1=1) so I could start to calibrate the groundwater elevation. When I ran my model for a year, I checked the groundwater elevation plot in the subcatchment with the LID control and the groundwater plot showed no recharge. When I took the LID control out of the subcatchment entirely, the groundwater plot showed recharge. When I set the LID control to occupy half of the subcatchment area, the groundwater plot showed recharge, but the recharge was at a greatly reduced rate as compared to 0% LID occupation. I would expect the 50% LID scenario to have higher (or at least as high) groundwater recharge given that LIDs infiltrate a much greater percentage of rainfall than regular land surface alone. When I ran a 24-hour simulation with different water table starting elevations, the LID summary showed the same depth of water infiltrated. It showed the same depth infiltrated even when the groundwater started above the BMP invert.

Looking at the LID Detailed Report, the "Bottom Infiltration" section is the same for all groundwater elevations EXCEPT when the starting elevation is equal to the surface elevation. When the GW is experiencing a loss (by either deep seepage or lateral flow), the values even out within a few days and then become the same. When the GW is experiencing no loss, the infiltration rate is the same EXCEPT when the starting elevation is equal to the surface elevation. In that circumstance, the infiltration rate is constant and small (0.03 in/hr). All result in the same amount of infiltration from the LID control according to the LID summary. (Note: I did not investigate the LID detailed report in 5.1.009.)

I know it's known that seepage from storage units and conduits is unaffected by the groundwater routine. (Are there plans to correct that, by the way?) Is the LID/groundwater interaction also a known error in SWMM and is there an easy workaround? Has anyone else had this problem? Or is it not an error at all and am I doing something wrong?

I have thought of the following as a potential workaround, though unideal: with the new flow routing options, I would think that sending all LID flow to a storage unit and calibrating a side orifice to match the underdrain and a bottom orifice to the approximate seepage rate. The bottom orifice goes to an outfall that dumps into a pervious subcatchment that never experiences any rainfall. The groundwater interacts with that outfall subcatchment and what runs off is water that's technically still in storage. That has implications for misrepresenting surface overflow, maybe. I don't know.

It's worth mentioning that when I ran the model in PCSWMM, the groundwater beneath the LID did recharge. I am unsure if this problem is isolated to EPA SWMM, but I am pretty sure PCSWMM is unaffected.

Lew Rossman

When the ability to route underdrain flow to separate locations was added to EPA SWMM 5.1.008 a lot of internal re-wiring had to be done to the LID routines to make this happen. As a result, it appears that a bug was introduced -- the LID infiltration was not being added on to the total subcatchment infiltration passed on to the groundwater routine like it was in earlier releases. If the LID infiltration was contributing to groundwater in your PCSWMM runs I assume it was because you were using the 5.1.007 (or earlier) engine in PCSWMM. Thanks for discovering this bug. I will work with Michael Tryby to try and get it fixed as soon as possible.