Case Study #2: Assessment of a proprietary underground structure for stormwater treatment

M. A. Wilson, J. S. Gulliver, O. Mohseni, and R. M. Hozalski at the University of Minnesota.

Hundreds of proprietary underground structures have been installed in the Twin Cities metropolitan area in recent years to improve the quality of stormwater runoff. One such placement is a V2B1 Model 4 by Environment21, LLC, installed in New Brighton, Minnesota, at the intersection of Rice Creek Road and Long Lake Road, as shown in figure 1. It receives stormwater runoff from a 4.2 acre, residential watershed that is approximately 55% vegetated and 45% impervious. Effluent from the V2B1 device ultimately reaches Long Lake because the lake receives effluent from the watershed.

The V2B1 Model 4 is a dual manhole system consisting of a 5 foot diameter swirl chamber and a 5 foot diameter floatables trap. Stormwater influent is introduced tangentially to the swirl chamber by the 15 inch PVC pipe, inducing a swirling motion inside the manhole. Relatively heavier particulates contained in the stormwater (sands, trash, etc.) settle out of suspension in the swirling chamber. Stormwater escapes the swirling chamber by overflowing an 18 inch diameter PVC standpipe in the middle of the manhole, where the water is conveyed to the floatables trap. The floatables trap manhole contains an underflow baffle wall with a 1 foot by 3 feet rectangular hole at its base. Buoyant material (hydrocarbons, cigarette butts, some organic matter, etc) that passes through the swirling chamber via the overflow standpipe are retained in the floatables trap since water must travel beneath the baffle wall to escape the system through a 15 inch PVC pipe. Downstream of the device, the effluent from the V2B1 discharges into a 36 inch reinforced concrete pipe (RCP), which eventually empties into Long Lake. There is an overall drop of 0.2 feet across the system, from the inlet invert to the outlet invert. The distance between pipe inverts and manhole inverts is approximately 4.5 feet in each treatment manhole. One access point is provided to the swirl chamber, and one access point on each side of the baffle wall in the floatables trap, as illustrated in figure 2.

The unit was designed to accommodate a maximum hydraulic flow rate equivalent to the 10-year event, with an intensity of 4.6 inches/hour, without flooding the street. According to calculations provided by Environment21, this discharge is 6.7 cfs, which serves as the capacity of the storm drain conveyance system around the device. The V2B1 is an in-line system with no bypass provided, meaning the device will receive all flows traveling through the system. However, even though all storm flows travel through the device, treatment is not intended to be provided above the water quality event, defined to be 0.8 inches of rainfall. A runoff coefficient of 0.46 was tabulated for the 4.2 acre watershed. According to calculations provided by Environment21, the water quality flowrate is 1.37 cfs, which corresponds to the maximum treatment rate for performance assessment.

Assessment Goals

The goals of this assessment were twofold:

  1. investigate the practicality of controlled field testing as an alternative to field monitoring
  2. to evaluate the sediment removal capability of the V2B1 Model 4 when subject to field testing with a wide range of sediment sizes and influent flow rates.

Another product of the assessment was a performance curve for the V2B1 in which removal efficiency is plotted versus a dimensionless parameter. This performance curve serves as a tool to reliably predict the removal performance for a wide range of V2B1 model sizes, influent flow rates, and pollutant size characteristics. The performance curve can also be used as a tool to accurately size a new stormwater treatment structure, given a target removal efficiency, a target particle size for removal, and a design flow rate.

Assessment Techniques

To meet the project goals, a new approach for stormwater facility assessment was developed and refined. Traditionally, stormwater facilities such as detention ponds, bioretention, underground structures, have been evaluated via field monitoring studies. Monitoring studies make use of sampling both upstream and downstream of a treatment facility such that improvement in water quality can be quantified. Upstream-downstream studies offer the advantage of evaluating removal performance of a facility when subject to the often wide variety of actual contaminants in a watershed of interest. However, due to the challenges of obtaining representative samples at both upstream and downstream sampling locations, data obtained are too general to specifically identify the range of performance and tend to have substantial uncertainty.

The concept of synthetic runoff testing for sedimentation devices is specific to performance as a sediment trap and avoids most of the uncertainties associated with monitoring. Using synthetic runoff, rather than actual storm events, utilizes water and sand that is artificially supplied to a clean device. At the completion of a test, personnel enter the device and remove the sediment retained during the test, allowing for a bulk solids analysis on a known quantity of delivered and retained sand. In addition to providing a more certain performance assessment, the synthetic runoff approach enables comparison of results for a particular device across different watersheds, climates, land uses (i.e., different pollutant loading), influent flow rates, and treatment unit size. This comparison can be accomplished by plotting the removal efficiency as the dependent variable versus the appropriate dimensionless parameter, as explained in the following paragraphs. Synthetic runoff testing is thus related to the performance of the device and not to the particular watershed. The runoff from the watershed can then be routed through the device using a computer simulation based on the characteristics of the watershed and the results of synthetic runoff testing.

Prospective sites from throughout the Twin Cities metropolitan area were identified, screened, and evaluated for field testing potential based on a variety of characteristics:

  1. location of out-of-vehicle traffic lanes for safety and traffic handling concerns
  2. proximity to a fire hydrant for use as a water source
  3. maximum treatment rate of the BMP device due to finite maximum discharges from hydrants
  4. device allowing for human access to treatment chamber sump for maintenance.

The system to be tested also needed to provide a suitable location within the storm drain system for flow rate measurement using a pre-calibrated weir and pressure transducer. Appropriate permits were obtained from governing agencies.

One of the sites chosen for field testing was the Environment21 V2B1 Model 4 device depicted in figures 1 and 2. Prior to beginning testing activities, the site required several preparation procedures:

  1. for real-time flow rate measurement, a pre-calibrated, 15–inch, circular weir and Campbell Scientific CR-10X pressure transducer (figure 3) were installed approximately 20 feet downstream of the floatables-trap manhole depicted in figure 2. The pressure transducer measured water depths, which, based on conduit geometry, were used to calculate flow areas and therefore discharge
  2. the V2B1 manholes were dewatered and several months worth of solids accumulation was removed with the assistance of vacuum trucks provided by the City of New Brighton
  3. a piping system was customized for the delivery of hydrant water as influent test water, using the hydrant’s 4 inch connection and a series of fittings, a 4-inch gate valve, and a 6-inch PVC pipe (figure 4)
  4. sand was previously sieved into three size fractions for use in each synthetic runoff event, with median sizes: 107 μm (ranging from 89 μm to 125 μm), 303 μm (ranging from 251 μm to 355 μm), and 545 μm (ranging from 500 μm to 589 μm), starting with F110 sand (d50~110 μm), AGSCO 40-70 sand (d50~225 μm), and AGSCO 35-50 sand (d50~425 μm) as supply
  5. an inflatable 15-inch diameter plug was secured from the City of New Brighton to seal off storm drainage upstream of the treatment system but downstream of the influent to prevent nuisance flows in the system from contaminating the controlled influent delivered to the V2B1 and to avoid controlled influent from leaving the test system prematurely.

The procedure for field testing the V2B1 Model 4 includes the following steps:

  1. establishing a safe work zone, following confined space entry regulations;
  2. installing and inflating with a portable air compressor the 15 inch rubber plug upstream of the V2B1 device to seal off the upstream reaches of the storm drain system (figure 5);
  3. connecting piping system from hydrant to influent injection point;
  4. flushing clean hydrant water through the system prior to initial device cleanout;
  5. dewatering the device with sump pumps and removing solids with a wet/dry vacuum cleaner;
  6. establishing an appropriate flow rate through the system using real-time level measurements from a pressure transducer and datalogger, and conditioning the flow with a gate valve on the hydrant. The datalogger recorded 60-second average levels and provided an updated readout every second when connected to a laptop computer loaded with Campbell Scientific’s PC200W software;
  7. introducing 10-15 kg of pre-sieved sand [equal parts of 107 μm, 303 μm, and 545 μm sands] to the influent hydrant water at 200 mg/L using a pre-calibrated sediment feeder;
  8. recording water temperature, mass of sediment delivered, and test duration;
  9. following a 20-minute period to allow sand particle settling, dewatering the device with sump pumps, and removing retained solids from each manhole separately with a wet/dry vacuum cleaner;
  10. oven drying and sieving the collected sediment into size fractions, and weighing each fraction of retained solids for comparison to the known quantity of each size fraction fed to the V2B1 during the test.

The data in step 10 above, divided by the known quantity of sand delivered to the V2B1 during the test, provided the removal efficiency of the device for each sand size fraction at a particular flow rate. Thus, each test produced three data points since three discrete sand size ranges were utilized. The testing protocol called for a device to be tested under four flow rate conditions in triplicate, at approximately 25%, 50%, 75%, and 100% of the maximum treatment rate (MTR), for a total of 12 tests. So under ideal test conditions, each device’s removal efficiency can be described by 36 data points.

A device’s removal efficiency can be plotted as a dependent variable against an appropriate dimensionless independent variable. The dimensionless parameter used as an independent variable was the Peclet Number (Pe), which is the ratio of advection to diffusion (Dhamotharan et al. 1981, Wilson et al. In Press). Advection is calculated as particle settling velocityVstimes a length scale L1. Diffusion can be simplified to flow rate Q divided by length scale L2. Putting advection and diffusion together yields Pe = (Vs*L1*L2)/Q, where L1 and L2 are taken to be a device’s treatment chamber diameter and settling depth.

As often as possible, the field team attempted to complete more than one test per day in order to maximize the effort in traveling to the site, setting up equipment, and preparing the device for testing, which were relatively constant ‘costs’ of testing whether 1 or 3 tests were performed. Construction activity adjacent to the stormwater quality test site presented difficulty with coordinating field testing. Additionally, a leaking swirl chamber was repaired to ensure proper hydraulics and system operation.

Assessment Results

At high Pe Vs (i.e., large particles and therefore high settling velocities), coupled with low flow rate Qs, a stormwater treatment device can be expected to be successful removing particles from an influent. If the Pe number was allowed to approach infinity (approximating a large detention pond or lake), very near 100% removal could be achieved. The data appear to exhibit this trend, but the required Pe to such removal is unknown. Conversely, at low Pe Vs (i.e., small particles and therefore low settling velocities), coupled with high flow rate Qs, a device can be expected to remove particles from influent with less success. This has been upheld in the results obtained, illustrated by the V2B1 performance curve depicted in figure 6.

The first several tests using the different particle sizes and relatively low flows indicated there was a problem carrying out tests with all of the sands designed for use during the experiment. Under low flow rates, the influent water velocity falls low enough such that it no longer can keep the largest sand particles in suspension for the entire distance from the injection point to the V2B1 (approximately 45 feet). Thus, heavier sands drop out of the water column and settle at the bottom of the pipe, a typical result of which is illustrated in figure 7. The experiment was modified such that the relatively low flow rates were increased (which therefore increased influent water velocities in the pipe) and the largest sand sizes removed from the mixture delivered to the device during these low-flow rate tests, producing a total of 30 data points in figure 6.

Conclusions and Recommendations

Understanding how devices perform under varying flow rates, sediment sizes, and treatment chamber sizes is important and helpful for consultants, local governments, and state agencies when selecting, designing, and evaluating stormwater treatment technologies for public infrastructure improvement projects. However, the effectiveness of proprietary underground stormwater treatment devices depends upon the settling velocity of influent solids (i.e., solid size and density) in addition to the size and design of the device. That Pe can be used to predict a device’s performance over a wide range of V2B1 model sizes, storm events, and pollutant size characteristics is possible because Pe relates two length scales and particle settling velocity to influent flow rate.

This research showed that controlled field tests are a practical, robust and accurate means of determining an underground device’s performance, based upon the solid size distribution and influent density, in addition to the water discharge and temperature. The results from this research have been successfully verified on three other devices in field tests and other devices in laboratory tests.

More specifically, these efforts have demonstrated the V2B1 capable of removing coarse solids at a relatively high rate (70% +), but is less efficient at removing fine sands (~32-48%). If the trend is projected to a lower Pe, one would expect that the V2B1 would be even less successful with finer particles such as silt, and remove few, if any, clay particles.

To predict performance and to determine appropriate device sizes, a suspended solid size distribution of typical runoff from the watershed is needed. The next goal is to develop a simple method of determining this size distribution of solids in stormwater runoff.


Dhamotharan, S., J. S. Gulliver, and H. G. Stefan. 1981. Unsteady One-Dimensional Settling of Suspended Sediment. Water Resources Research 17:1125-1132.

Wilson, M., J. Gulliver, O. Mohseni, and R. Hozalski. In Press. Controlled field study of underground proprietary sediment trapping devices. Master of Science. University of Minnesota, Minneapolis.

Return to Case Studies.