Chapter 19 Consistent and Uniform Method for the Evaluation of Stormwater BMPs

As a design parameter, the first-flush concept has been the most practiced, and most questioned, criterion used by engineers. Quantifying runoff remains a daunting challenge due to the extreme variability in rainfall events, pollutant loading, and surface conditions among other factors. Advocates of using the first flush concept as a design parameter have typically attempted to quantify it as the first0.5inch(13mm)orinch(25mm)orotherdepthofprecipitation. Othershave preferred to quantify it as depth of runoff as opposed to precipitation. The following discussion will show that rainfall intensity, not depth, may be the most appropriate design parameter for evaluating runoff quality. Although the first-flush concept holds some legitimacy, many studies have shown that


.I Runoff Quality
As a design parameter, the first-flush concept has been the most practiced, and most questioned, criterion used by engineers.Quantifying runoff remains a daunting challenge due to the extreme variability in rainfall events, pollutant loading, and surface conditions among other factors.Advocates of using the first flush concept as a design parameter have typically attempted to quantify it as the first0.5inch(13mm)orinch(25mm)orotherdepthofprecipitation.Othershave preferred to quantify it as depth of runoff as opposed to precipitation.
The following discussion will show that rainfall intensity, not depth, may be the most appropriate design parameter for evaluating runoff quality.Although the first-flush concept holds some legitimacy, many studies have shown that "concentration profiles for particle-associated pollutants often display a discharge pattern more complex than a simple 'first-flush'" (University of Texas, 1993).Studies by Hoffinan et al. (1985) revealed that runoff concentrations for hydrocarbons, lead, and suspended solids "generally followed the trend of rainfall intensity."Harrison and Wilson (1985) confirmed these findings by reporting that "the temporal variation of concentrations of particle-associated pollutants was more complex [than a first-flush] and related to rainfall intensity and the flushing of sediment through drainage systems." A particulate washoff model presented by Sartor and Boyd (1972)  The proportionality constant, k, is a subjective variable chosen by the model user.For the model to produce reliable results, the proportionality constant should be calibrated at the intended site.Sartor and Boyd maintain that the proportionality constant is only slightly dependent on street texture and condition and independent of rain intensity and particle size.Their recommended k-value of 0.18/mm is intended to produce 90% washoff of the sediment load from the street surface in one hour during a 13 mmlhr rainfall event.Many analysts have criticized their value for k claiming that it exaggerates washoff quantities.Vladimir Novotny has proposed much smaller values for k, specifically 0.0261 rom for fine particles (<45 micron) and 0.0 lImm for medium sized particles (100 to 250 micron).Studies by Jewell et al. (1980) and Alley (1981) also found considerable variation for k values and both recommend local calibration (Pitt,19l:H).
While the model incorporates both rainfall intensity and duration into its estimation, it doesn't directly relate buildup and washoff of sediments to rainfall intensity.To exclusively account for the effect of rainfaU intensity on washoff quantities, Novotny & Chesters (1981) developed the so-caned availability factor (A).The purpose of the factor is to isolate the quantity of sediments available for washofI given the relative energy of rainfall.The availability factor is given by: A = 0.057 + 0.04 (r )1.1 (19.2)

Analysis
where: r = rainfall intensity (mm/hr) The product of the availability factor and the initial sediment loading amount (NO> yields the total available sediment load susceptible to washoff at a given rainfall intensity.The Sartor & Boyd model can then be written: For all intensities greater than or equal to 18 mm/hr, the availability factor is equal to 1.0 (hence, all sediments are available to washoft).Popular runoff models STORM and SWMM both utilize the factor for calculating the available sediment loads (Pitt, 1987).

Analysis
The following analysis was developed so that direct comparisons can be made between stormwater quality best management practices (BMPs).Modeling two or more stormwater quality BMPs under similar runoff conditions and comparing their perfOJ"Q1ance will allow engineers to make appropriate decisions with respect to a site's stormwatertreatmentneeds.Requirements fortheanalysis include a description of a rainfall event, a classification of surface constituents, alocallycalibratedorassumedvalueofk,andperformancedataforaspecific BMP.

Selecting a Rainfall Event
The first step to applying the Sartor & Boyd model is to describe a rainfall event in terms of intensity and duration (i.e.drawahyetograph).Using a regional design storm (i.e.USDA Soil Conservation Service (SCS) Storm) for the site being modeled would be one alternative; using an observed storm that typifies the desired parameters is another alternative and the approach adopted for this discussion.Storms that maintain 6-month storm intensities or greater are recommended because they represent a threshold upon which sediment transport becomes significant.This analysis will use an observed rainfall event that occurred on January 20, 1995 in Southeast Cumberland County, Maine.This storm was a 6-month storm in terms of peak intensities (see Figure 19.1).

Surface Constituents
To evaluate how a BMP would perform at a certain site, the surface constituents must be descnood in both quantitative and qualitative terms.The Sartor & Boyd model requires the sediment load (NO> be quantified in terms of 16,---------------------------------------- mass per linear measure.Since this analysis will focus only on the percentage of sediment mobilized, the selection of No (e.g.initial sediment load) is inconsequential.The composition or quality of the initial sediment load must be described, preferably in the form of a particle gradation curve.Countless particle size distributions have been conducted on sediments from roadways, parking lots and other areas subjected to vehicular traffic and stormwater runoff.Physical properties of sediment unquestionably vary from site to site.For accurate results, a particle gradation curve should be performed at the site being modeled.For qualifying the sediment load in this analysis, we will use a particle gradation curve produced by Pitt (1979) for the San Francisco Bay area (Figure 19.2).

Sediment Transport
Sediment transport certainly varies widely.The variation in the two curves demonstrates how influential the proportionality constant is to the model.Observing the curve formed when k is equal to O.026/mm, one can see thatthe storm washed off a total of33.3% of the available sediment from the surface.The model indicates that intensities were not significant enough to washoffthe remaining 66.6%.When k was set equal to 0.1 03/nun, the model predicted 87% washoff of available sediments.

Evaluating a BMP
The final phase of the analysis is to select a BMP and establish removal efficiencies for uniform particle sizes for a range of flow rates.The Vortechs ™ Storm water Treatment System will be used to demonstrate the application of this analysis.The Vortechs System (see Figure 19.4), designed and manufactured by Vortechnics, Inc., provides physical treatment for stonnwater runoff.Attributes of the system include a unique swirl concentrator, which promotes particle settling, accompanied by a surge control feature to contain floatables and maintain quiescent flow conditions.
surface area.Almost any BMP operating rate can be represented in similar fashion.Figme 19.5 shows performance data for the Vortechs System.
To determine the level of performance the system will achieve given the nature of street sediment and rainfall, it is presumed that finer sediments are mobilized before larger sediments during a rainfall event The supposition is that a particle with less mass requires less force to be mobilized than a larger particle.The incremental intensity growth typical of most rainfall events suggests that finer sediments are mobilized during the early stages of an event and larger sediments are mobilized during subsequent, more intense stages.
To establish the flow rate at which the Vortechs System was performing during each interval, it's necessary to correlate rainfall intensity to flow rate and operating rate.If this was a real site and the acreage and land description was provided, it might be appropriate to use the rational method (ifless than 100 acres

Analysis
or 50 ha) to determine the flow rate to the system.The flow rate could then be correlated to the system's operating rate.Vortecbnics' sizing criteria state that a 2-month storm (approximate intensity of 12 mmlhr, for SE Cumberland County, Maine) should operate a Vortechs System at no greater than 24 gpm per sq.ft.
It follows that if a 12 mm/hr rainfall intensity operates the system at 24 gpmIsq.ft., then a 1 mm/hr rainfall intensity will operate it at no greater than 2 gpmIsq.ft and a 2 mm/hr rainfall at 4 gpmIsq.ft. and so on.(A complete description of Vortechs System sizing criteria is available from Vortechnics Technical Bulletin No.3).

Predicting BMP Performance
Evaluating the performance of the Vortechs System will entail using Figures 19.2, 19.3 and 19.5 in combination.This evaluation will use the curve basedonNovotny'sk-valueofO.0261mm.FromthecurveinFigureI9.3,observe that approximately 3% of the total available sediment had washed from the surface after 1 hr of rainfall.On the premise that finer particles are mobilized first, that 3% is presumed to represent the finest fraction of the initial sediment load.Therefore from Figure 19.2, about 3% of the sediment load consists of particles about 30-microns or smaller and this is the portion that the model predicts will be mobilized in that :first hour.The average intensity during that interval was 4 mmlhr and the operating rate was 8 gpmIsq.ft.From Figure 19.5 we see that the Vortechs System removes 800/0 of the 30-micron sized particles at 8 gpmIsq.ft.Therefore, at the I-hour mark, the system was operating at 80010 efficiency.
In view of estimates being on the high end of the range (Le.using the maximum particle size), results will be increasingly accurate with smaller intervals (i.e. 5 to 15 min).Figure 19.6 shows how the Vortechs System performed in the 6-month event evaluating conditions every hour.

Conclusion
With a growing number ofBMPs and manufactured products emerging for stormwater quality management, there exists an ever-greater need for a common basis of comparison of the various alternatives.
The approach described in this chapter is an attempt to provide such a basis.The key element needed to implement this approach would be the development of a set of curves such as those shown in Figure 19.5.The units of operating rate may differ.For example, the units might be gpm per sq.ft.(as applied to the Vortechs Systems), detention time, or flow-through velocity.Another key factor is that rainfall intensity must be correlated to operating rate.
With such a set of completely descriptive curves, comparing the curves for one BMP with the curves for another will be straightforward.Some BMPs may exhibit efficiencies that approach zero or even drop below zero at higher rainfall intensities and operating rates.Furthermore, once a set of curves for each BMP is developed, a single curve for each one could be developed, by applying the same model that was presented here in developing Figure 19.6.
The authors believe that this model will make a very significant contribution to the field of storm water quality management.Currently, that field is notoriously lacking in any standards, design parameters or criteria that should be met and even more lacking in consistent ways to evaluate the measures that are supposed to meet what standards do exist This model is intended to make evaluations of both conventional and innovative measures much more uniform and consistent To the extent that it succeeds in doing so, it will further the cause of developing stormwater quality standards and design parameters that are achievable.
Figure 19.2Particle size distribution of street refuse before sweeping (pitt, 1979).

Figure 19 . 3
Figure 19.3Sediment washoff for a 6-month storm in southeast Cumberland County, Maine.