Use of TMDL Credits for BMP Comparisons

Florida Department of Environmental Protection (FDEP) issued total maximum daily load (TMDL) mandates in March 2009 for the Indian River Lagoon (IRL), a bay in east central Florida. (FDEP 2009). The TMDL requires communities to reduce nutrient loadings in their stormwater runoff. FDEP is currently undertaking the implementation stage of the TMDL program, called a Basin Management Action Plan (BMAP), which is enforced through National Pollutant Discharge Elimination System (NPDES) municipal separate storm sewer system (MS4) permits for entities in the IRL. The IRL drainage basin encompasses 16 cities, three counties, four military bases, eight water control districts, and two Florida Department of Transportation (FDOT) districts.

Impairment in the IRL was measured by loss of seagrass, as benchmarked from a combination of aerial photography and seagrass transects between 1945 and 2000 (FDEP 2011). The HSPF model was used to correlate nutrient loadings to seagrass extents. It was used to model the IRL as a whole, but was too coarse to differentiate individual MS4 basins for regulatory purposes.

Nonpoint-source loads from stormwater runoff were the primary source of pollutants identified in the TMDL. Point-source loads were not found to be significant because wastewater treatment plant (WWTP) discharges to the IRL have been banned since 1990. Atmospheric discharges were also found to be insignificant (FDEP 2009).

Figure 1. Project cost versus annual TP removed by type

Figure 2. TN and TP removal costs for four BMP types

For regulatory and compliance purposes, a customized version of the GIS-based Pollutant Load Simulation Model (PLSM) was developed that allowed discrete analysis of stormwater loadings from each sub-basin, as well as reductions from existing and proposed best management practices (BMPs) throughout the lagoon. PLSM generated existing pollutant loads from multiple, spatially distributed inputs such as land use, soil types, hydrologic boundaries, rainfall, runoff coefficients, event mean concentrations (ECMs), and BMP type.

The BMAP requires a lagoon-wide reduction over a 15-year period of 380,239 pounds per year of total nitrogen (TN) and 84,840 pounds per year of total phosphorus (TP), which equates to 45% and 58% reductions of TN and TP, respectively. Reductions are measured as pounds of pollutant per year and are allocated among MS4s according to their proportioned individual baseline loadings.

New development permitting regulations in the IRL require that post-development pollutant loads match predevelopment pollutant loads. Therefore, new development is assumed to have no net impact on impaired waters and is invisible to the TMDL process. Reduction compliance can be achieved through implementation of various structural, maintenance, and public education programs customized for each MS4. The majority of reductions will be achieved through structural retrofit projects in older developments that have few or no BMPs for pollution control.

Study Objective

Most MS4s in the IRL will struggle to attain BMAP compliance due to financial limitations and shortage of available land to construct retrofit projects of significant size, although some MS4s may have a surplus of pollutant reduction credits.

Stormwater Solutions was tasked with development of a TMDL Credit Valuation that could be used for trading pollutant removal credits among entities. Credit purchases could be used by entities to avoid costs and resources required for the process of land acquisition, construction, and maintenance of retrofit projects within their jurisdiction. Costs and pollutant removal effectiveness of 75 BMPs in the IRL basin were analyzed, and development of a common metric for credit pricing established the value of TMDL credits.

BMP Comparison Strategy

Numerous BMP comparisons and performance metrics have been developed in the past by cumulating records of historical projects in diverse locations, over varying timeframes, with mixed pollutant calculation methods. This type of data mining leads to weak data variable control, resulting in inaccurate comparison results.

Datasets were controlled in this analysis with the goal of reducing variability as much as possible from pollutant load calculations, BMP efficiencies, cost estimates, and watershed uniqueness. A strategy was devised for comparing BMP cost and performance within the context of TMDL allocations and compliance. The strategy utilized the following principles:

  • All comparable BMPs were located in the IRL.
  • BMPs were on FDEP’s list of BMP types that qualified for TDML credits.
  • The same pollutant load model was used for all loading calculations.
  • The same criteria were used for selection of all model variables.
  • Consistent pollutant removal efficiencies were used for all BMPs.
  • Cost estimates were developed using current construction costs.
  • A consistent metric was used for comparison of BMP performance.

FDEP controls TMDL compliance in the BMAP process by performing all pollutant loading and BMP modeling with the PLSM model. Three TMDL master plans were recently developed by Stormwater Solutions for Indian Harbour Beach, Satellite Beach, and parts of Brevard County (England 2011a-c), all within the IRL. These were strictly water-quality master plans developed without traditional hydrologic or hydraulic modeling. Stormwater Solutions used FDEP’s PLSM model to develop the master plans, thereby ensuring pollutant calculation consistency and minimizing data variability in this study. From these master plans, 75 municipal retrofit projects specifically designed for TMDL compliance, rather than flood reduction, were selected for comparison and analysis.

Types of BMPs

For TMDL treatment purposes, FDEP recognizes only a limited number of BMPs that have FDEP-documented removal efficiencies. Table 1 shows the BMPs and associated removal efficiencies that may be used to calculate load reductions for BMAP compliance. Provisional BMP values will be adjusted as additional research is compiled.

The majority of soils throughout the master plan areas are Type A or A/D, with low, flat topography. Groundwater elevations vary from 1 to 4 feet below surface elevations.

BMPs types recommended by Stormwater Solutions in local master plans to achieve significant levels of nutrient removal were limited to wet detention ponds (WETP), dry retention ponds or LID swales (DRY), exfiltration trenches (EXF), and floating vegetated islands (FWET). Vault-type BMPs provide nutrient reductions principally through screening of leaves and organic debris (England and Smith 2010). Within the BRL drainage basin, there was minimal tree canopy coverage in the watersheds to provide leaf debris. Therefore, vault-type BMPs were not recommended in the master plans.

Load Calculations

In Florida 87.8% of rainfall events have 1 inch or less of precipitation, and 75.1% of events produce less than 0.5 inch of rain. (Harper 2007). The majority of rainfall events result in low runoff volume, especially in soils with high permeability. In the IRL, annual rainfall was spatially distributed over the basin, ranging from 46.66 to 54.65 inches per year. For pollutant loadings and BMP removal efficiency calculations, FDEP uses the metric of pounds of pollutant generated on an annual basis to normalize the frequency distribution of many small storms that generate little or no runoff volume or pollutant loads. TMDL pollutant load models do not use the same rainfall distributions and frequencies that are required for new development stormwater permitting.

The BMPs selected for comparison in this study were modeled with the same PLSM model used for IRL BMAP regulation. Individual subbasins were keyholed out of the master model, resulting in consistent variables and calculation methods for all pollutant loadings. All BMP removal efficiencies were calculated using the methods from Table 1.

BMP Costs

Estimation of BMP cost introduces potential variability to BMP comparisons. Ten different engineers will give 10 different costs estimates for a project. Unit prices vary with time, economic conditions, and contractors. In 2010, the author used consistent unit prices to generate each of the 75 project cost estimates in this study.

BMP costs for retrofit projects are highly affected by land acquisition costs. Land values can be higher than BMP construction costs and can vary widely due to many variables of zoning and location. Of the 75 projects selected, 13 had land acquisition components. To normalize cost variability, land costs were removed from all BMP cost estimates in this report.

A life-cycle cost analysis with net present values is often used for long-term BMP cost estimates. Life cycles vary widely by type of BMP. A wet pond might last 20 years before reconstruction, exfiltration trenches last five to 10 years, and floating vegetated islands are replaced every year.

In the context of municipal BMP construction and maintenance, two different budget sources should be recognized. A capital improvement program (CIP) budget with revenues from stormwater utility fees, grants, and bonds is used for construction of BMPs. Long-term maintenance is generally undertaken using general funds and other sources. Use of a life-cycle cost analysis to combine short-term construction costs with long-term maintenance costs is theoretically accurate, but not useful for most MS4s because of differing funding sources. A poll of local communities indicated that their main consideration for TMDL credit pricing would be the initial construction cost, rather than long-term maintenance costs. Accordingly, a net present value analysis was not undertaken to factor long-term BMP maintenance into overall costs. Cost estimates in this report included 15% engineering and permitting, 8% surveying, 20% contingency, and 3% operation and maintenance (O&M) costs.

Analysis

BMP cost-comparison studies often use a common denominator of acres treated to meet a regulatory goal that has a presumptive criteria of 80% removal or a specified pollutant concentration. While this type of comparison works well for new development scenarios, available land limitations may prevent a retrofit pond from meeting the size required for new development permitting. In these situations the BMP will not provide a theoretical 80% removal. For TMDL purposes of retrofitting existing development, the number of acres treated is not a valid metric because the goal for TMDL compliance is not to design a BMP to treat a certain number of acres. Rather, the goal is to maximize the number of pounds of pollutant than can be removed at a specific site given the limitations of BMP selection and limited land size, regardless of the drainage basin size.

Figure 3. Area-weighted TP removal by BMP type

Figure 4. Interval plot of cost per pound/acre by BMP type

Annual pollutant loads removed for each parameter, PLRp, were calculated with the following equation:

PLRp = PLp x RE

where PLp is the pollutant load calculated from the PLSM model and RE is the removal efficiency from Table 1.

BMP Comparisons

The size of drainage basins treated ranged from 0.88 to 549.63 acres, with a median size of 20.82 acres. Estimated project costs (without land) varied from $1,779 to $1,431,480, providing a broad range of project conditions and scales for analysis. Table 2 shows that sample sizes for BMP types ranged from 12 for DRY to 28 for FWET.

In an initial analysis, all BMPs were combined into one dataset. Standard deviations of $4,110 per pound and $34,701 per pound for TN and TP indicated lower than desired statistical correlation (r2 = 0.44 for TN and TP) for use of one dataset for all BMPs. When BMPs were consolidated by the treatment types, statistical linear regressions showed stronger correlations between cost and pounds of pollutant removed, although there was still high variability of costs, especially with wet ponds. Figure 1 shows cost versus TP Removal. TN had similar correlations. EXFs had an r2 = 0.79, and WETPs had an r2 = 0.67.

To minimize data variability and meet the TMDL goal of removing the most pollutants for the least cost, metrics of cost per pound of TN and TP removed (without land acquisition cost) were chosen for the TMDL credit valuation in this study. Since treatments types are important indicators of both cost and pollutant removal efficiency, the remaining analyses in this study were conducted by treatment type.

Table 3 shows a summary and comparison of BMP types and costs. Due to high variability of the datasets, mean values for cost per pound removed were unrealistically high. Therefore, median values rather than mean values are used in this analysis. DRY had the lowest cost per pound of TN and TP removed. FWET, WETP, and EXF followed in ascending costs per pound.

Note that WETPs were the only BMP that could singly treat large drainage basins. Correspondingly, WETP pollutant removals were an order of magnitude higher than those of other BMPs, demonstrating that many smaller BMPs would be necessary to achieve the same pollutant removal as one regional WETP.

EXF cost per pound removed for TP was sometimes an order of magnitude higher than those for other BMPs (Figure 2). Investigation of EXF construction costs revealed high costs due to pavement replacement and numerous utility adjustments associated with this type of BMP. Exfiltration trenches also had low runoff storage volume that provided limited treatment ability, resulting in very high cost per pound removed values. For this reason, EXFs are generally used as a last resort when there is no other BMP alternative. Mean cost per pound removed with and without EXF also demonstrated how EXF BMPs skewed the overall dataset using all BMPs. Therefore, EXF was not considered to be a representative BMP type for credit pricing in this report. Mean costs per pound of TN and TP removed for all BMPs (without EXF) of $528 and $2,749 are the recommended values of TMDL credits in the IRL.

TP removal costs were consistently higher than TN removal costs (Figure 2), making TP the more relevant metric to use in the IRL where both parameters had allocations.

A linear regression analysis indicated there was very low correlation between cost and any single variable, but taking treated area size into account, much better correlation between area-weighted pollutant removal and cost was encountered. In the comparison between BMPs shown in Figure 3, dry ponds had the highest area-weighted TP removal effectiveness per pound/acre, indicating DRY was the most efficient BMP, followed by WETP, EXF, and FWET. Note that costs are not factored into this rating.

Considering the many variables involved with pollutant loadings and BMP treatment, one can understand that why there is no simple correlation between only the two variables of cost and pounds removed. A multivariate analysis showed a strong correlation between cost, drainage area treated, and pounds of TN or TP removed. An interval plot of these three variables by BMP type showed sufficient correlation to give the 95% confidence intervals shown in Figure 4. DRY had very tight data and lowest mean cost of $16,560 per pound of TP removed per acre treated, followed by $700,864 for FWET, $748,909 for EXF, and $1,060,540 for WETP. The mean values of cost per pound of TN removed per acre treated were $3,630 for DRY, $63,196 for FWET, $112,819 for EXF, and $427,695 for WETP.

While the wide disparity of these numbers makes them difficult to be useful in an absolute sense, these parameters can be utilized in prioritizing a list of BMPs for cost efficiency. FDEP uses this same multivariable comparison in ranking applications for TMDL Water Quality Restoration grants.

Conclusions

The goal of this study was to calculate a valuation of TMDL credits for MS4 trading purposes. A metric of costs per pound of pollutant removed was chosen to allow flexibility for utilization with any pollutant listed for impairment. The parameters used in this analysis were TN and TP as required in the IRL TMDL. Analysis showed land costs unreasonably skewed the dataset, so land costs were removed from BMP costing.

Initially the four types of BMPs investigated were wet detention ponds, dry retention ponds, floating vegetated wetlands, and exfiltration trenches. Exfiltration trench costs were found to be significantly higher than other BMP types, exfiltration trenches were not recommended as a comparable BMP type for this valuation analysis.

Due to the high variability of stormwater loadings and BMP costs, use of mean cost per pound removed gave unreasonable values. Therefore, median values of cost per pound removed were used for data analysis. For TMDL credit valuation purposes, costs per pound of pollutant removed were $528 for TN and $2,749 for TP. This metric indicated a ranking of costs per pound removed from lowest to highest of DRY, WETP, FWET, and EXF.

There was a stronger statistical correlation between cost and acres treated than between cost and pounds removed. A multivariate analysis indicated that the metric of cost per pound removed per acre treated was highly correlated and recommended for ranking different types of BMPs for the highest cost effectiveness. The results of this analysis showed the ranking of BMP cost effectiveness from highest to lowest was DRY, FWET, EXF, and WETP.

Both metrics showed that dry ponds were the lowest-cost BMP. The differing metrics gave different rankings for the other three BMPs, demonstrating the highly variable nature of stormwater treatment. Cost per pound removed per acre treated is the more statistically accurate metric. Other factors, such as number of pounds to be removed for TMDL compliance (the need for large versus small projects), land availability, and politics, also play important roles in BMP selection.

About the Author

Gordon England and Claudia Listopad

Gordon England, P.E., D.WRE, is with Applied Sciences Consulting Inc. Claudia Listopad, Ph.D., is with Applied Ecology Inc.