Temperature effects on functionalized filter media for nutrient removal in stormwater treatment



Attempts to achieve better removal efficiencies of nutrients in stormwater treatment by using aggregates (e.g., mixes of tire crumbs, sawdust, sand, clay, zeolite, sulfur, or limestone) in filter media, has been a common practice in green infrastructures. These material mixes mainly promote the adsorption/absorption and precipitation of orthophosphate in the physicochemical process step and the transformation of ammonia, nitrite, and nitrate via oxidation and reduction reactions in the microbiological stage. Some processes, however, are known to be active only in a limited temperature range and variations in filtration kinetics of the sorption media are still unknown. This article aims to explore the filtration kinetics of selected filter media mixes for nutrient removal at various temperatures. With the basic understanding gained in material characterization, a laboratory column study was conducted to simulate the conditions in saturated media. A kinetics study of this kind allowed a comparison of a natural soil with soil augmentations in terms of nutrient removal within a range of the initial concentrations and temperatures. The temperatures (28, 23, and 10°C) were selected to reflect the normal temperature variations in subtropical regions. Significant differences of nutrient removal efficiencies associated with these temperatures were statistically confirmed by ANOVA analyses. The temperature correction factor model was finally applied to address the impact on treatment processes due to the seasonal temperature variations. © 2010 American Institute of Chemical Engineers Environ Prog, 2011.


The use of sorption media mixes in biofiltration practices may achieve a plethora of improvements for nutrient removal in stormwater management [1]. Some physicochemical processes such as the adsorption, absorption, ion exchange, and precipitation reactions are actually intertwined with microbiological processes of nitrification and dentrification in these porous media. Regardless of the geographical locations, total nitrogen (TN) in the stormwater runoff would normally consist of ammonia (NH3), nitrite (NOmath image), nitrate (NOmath image), and organic nitrogen (ON), whereas total phosphorus (TP) would mainly consist of particulate phosphorus (PP), soluble unreactive phosphorus (SUP), and soluble reactive phosphorus [SRP; approximately soluble inorganic orthophosphate (PO4)]. Before 1995, research studies in this area were primarily concerned with the sand filter method. For this reason, different types of sand filter methods had been developed such as: (1) the Washington D.C. sand filter method, (2) the Delaware sand filter design, and (3) the Austin sand filter [2]. The removal efficiency of the Delaware sand filter is 70.20% of Total Suspended Solids (TSS), 71.10% of Total Phosphorus (TP), 67.0% ammonia nitrogen (NH3-N), and 59.90% of Total Kjeldahl nitrogen (TKN), [3]. Several studies showed that the removal of ammonia, nitrite, nitrate, and phosphorus can be enhanced by mixing different sorption media, such as sawdust, tire crumbs, sand, clay, zeolite, sulfur, and/or limestone, etc., with natural soil [4–8].

Filtration media mixes in our study are defined as multifunctional materials or functionalized sorption media that may be used in both natural systems and built environments to improve the existing processes for nutrient removal. Such media mixes have “green” implications because of the inclusion of some recycled materials, such as tire crumb and sawdust, as part of the recipe to promote the treatment efficiencies [1]. With the aid of such green sorption media, nutrients in water bodies can be reduced or even mostly removed by enhanced absorption/adsorption, nitrification/denitrification, and other chemical reactions, such as precipitation and ion exchange. However, many engineering and natural factors, including the initial concentrations, loading rate, material properties, pH values in the water body, and ambient temperatures, might affect the ultimate success of nutrient removal in sorption media. Temperature has a significant effect on nitrification that must be taken into consideration during design [9]. In general, colder temperatures require lower hydraulic loading rates in attached-growth systems due to slower growth rates of nitrifying bacteria. Yet nitrifiers can grow best in a temperature range of 35–42°C and denitrifiers can work well in a range of 10–25°C [9, 10]. Nitrification and denitrification are also heavily dependent on stormwater pH because values <5.7 will inhibit nitrification [11]. The optimal pH range for nitrification and denitrification is between 7.5–8 and 7–8, respectively [10].

Since how temperature changes could affect the nutrient removal efficiencies in green sorption media remains as yet unknown in the literature, this article aims to explore how the filtration kinetics of selected filter media for nutrient removal could be affected by various temperatures. It starts with a material characterization of a selected sorption media, followed by a laboratory column study that was conducted to simulate a stormwater treatment unit with saturated media conditions similar to those found in stormwater retention ponds. The actual filtration kinetics study systematically compared the nutrient removal efficiencies between the engineered cases and the control case (i.e., a natural soil) based on a range of the initial nutrient concentrations associated with three different temperatures (i.e., 28, 23, and 10°C). The temperatures selected to reflect the seasonal variations a year round in subtropical regions. The temperature correction factor model was developed to address such temperature effects on nutrient removal.


Material Characterization

We selected two media recipes, media mix 1 consisting of 50% fine sand, 30% tire crumb, 20% sawdust by weight, and media mix 2 consisting of 50% fine sand, 25% sawdust, 15% tire crumb, 10% limestone by weight. Sawdust, tire crumb, limestone, and sand were purchased from different local vendors. They were mixed by weight based on our prescribed recipe. Hunter's Trace soil was collected at the Hunter Trace stormwater retention pond at Ocala, Florida.

ASTM Standard Practices were used to determine the density, void ratio, porosity, specific gravity, and permeability. First of all, the ASTM D-421-85 Standard Practice for Dry Preparation of Soil Samples for Particle-Size Analysis and Determination of Soil Constants was used to generate the particle size distribution. The first step in the sieve analysis was to determine the mass (grams) of dry sample tested. The sieves were prepared by stacking them in an increasing order using sieve numbers 4, 10, 20, 40, 60, 100, 140, 200, 230, and 270, from top to bottom, respectively. A bottom pan was placed under the stack of sieves. The sample was then poured into the stack of sieves and covered with a sieve cover. A sieve shaker shook the sieves for approximately 10 min. After the sieve shaker had stopped, the stack of sieves was removed. The amount of soil retained on each sieve was weighed, starting from the top sieve (No. 4) to the bottom sieve (No. 270), and the bottom pan to draw the gradation curve.

The specific gravity was measured using the ASTM D-854-92 Standard Test Method for Specific Gravity of Soils. The measured volume of the media was 100 g. The pycnometer was a volumetric flask having the capacity of 1000 mL. The coefficient of permeability was found using the falling head test method. The preparation of the soil specimen follows the same procedure as the ASTM Standard D 2434-68 for the constant head method. After the specimen was saturated, any air bubbles within the tubing were removed. The time for the water to flow from the two selected heads, h1 to h2, was measured. Several trials were run and averaged. Then the permeability was converted to a test temperature of water at 20°C. Surface areas of sorption media were measured by a certified laboratory (Quantachrome Instruments) using “Autosorb Station.”

Reaction Kinetics

To calculate the rate of a reaction, it is assumed that the simplified form of an equation is used for the overall reaction. Equation 1 is a general version of the zero, first, or higher order rate equations.

equation image(1)

Rates of reactions can be influenced by factors such as temperature, concentration, and the presence of a catalyst. To address the temperature impact on reaction kinetics, the temperature correction factor model relates the rate constant to temperature (Eq. 2). The variables in the model are defined as rate constant at temperature T (kT), the rate constant at 20°C (k20), the temperature correction factor (θ), and temperature in °C (T). The model is generally applied to biological processes that have low temperature dependence. A temperature range of around 20°C was used as the reference temperature in support of three kinetic models that were considered associated with differing temperatures in the subsequent column tests.

equation image(2)

Design and Setup of Column Tests

The treatment for nutrient removal using sorption media may be flexibly arranged in stormwater retention/detention ponds, groundwater remediation sites, and fully controlled engineered reactors for handling domestic and industrial wastewater effluents. In the treatment unit, stormwater may be supplied at the top of the media and flows downward by percolation. In this case, if the hydraulic design allows the top layer of sorption media to maintain an aerobic (unsaturated) environment while the bottom layer maintains an anaerobic (saturated) environment, the influent can most likely experience both physicochemical and microbiological processes successively. It is expected that when the influent reaches the layer unsaturated, ammonification and nitrification will be active causing ammonia adsorbed on the surface of sorption media to undergo nitrification.

According to this interpretation, the columns were designed to simulate field environments with saturated conditions. By using three columns, it is possible to test the performance of two mixes as compared with a control. The first control column is filled with natural soil collected from the Hunters Trace stormwater pond in Ocala, Florida. The soil was sun dried and sieved with a no. 10 sieve to remove vegetation, rocks, and large particles. The two media recipes selected are media mix 1 consisting of 50% fine sand, 30% tire crumb, 20% sawdust by weight and media mix 2 consisting of 50% fine sand, 25% sawdust, 15% tire crumb, 10% limestone by weight based on a multicriteria decision making (MCDM) analysis [12].

The clear Plexiglas columns have an inside diameter of 14.7 cm (5.8 in.) and are 91.4 cm (3.0 ft.) in length. The bottom of each column contains a filter with 3 in. of fish tank rocks to prevent the media and soil mixes from exiting the columns and clogging the tubes. These columns were secured onto a constructed wooden frame with straps. Two ports were installed in each column to take core samples at different heights and hydraulic retention times (HRT). In addition to the ports, the water was sampled from the bottom of each column. The gaps surrounding the ports were sealed with Rectorsell 5 and Plumbing Amazing Goop. One peristaltic pump with peristaltic tubing was used to pump water from three 5 gallon reservoirs into each of the columns. A fourth reservoir was used to maintain saturated conditions by connecting the column outlet tubing at a height equal to the desired water level inside the columns. Figure 1 shows the schematic of the column setup.

Figure 1.

The schematic of column design.

The HRT for the columns was calculated by using the following equations in a MathCAD file so that any changes in parameters could be quickly adjusted. The contributing parameters include porosity (n), column inside diameter (dcolumn), and the height above sampling port (h). First, the volume (Vmedia(h)) was calculated using Eq. 3.

equation image(3)

Then, using this volume and the desired HRT at the bottom of the column, the flow rate (Q) for the pump was calculated by Eq. 4.

equation image(4)

Lastly, by using this flow rate, the HRT for each port (θ) can be calculated by Eq. 5.

equation image(5)

The columns were placed in a room in which a constant temperature for two of the temperatures could be set and a refrigerator for the lowest temperature. This design allowed the testing of the media mixes at various constant temperatures to determine the temperature impacts on filtration kinetics.

Column Analysis

The pond water collected from a stormwater retention pond was first augmented with KNO3 and HK2PO4 to preselected ranges and then pumped from each of the reservoirs through the columns. The concentration ranges were employed to validate the kinetics. Water samples were taken from three locations of each column; Port 1, Port 2, and Port 3 (i.e., the bottom of column). These samples were then tested for variations of nitrogen and phosphorus. A summary of these methods is listed in Table 1.

Table 1. Column study water quality parameters and methods
  • *

    Note: In some cases samples were diluted with DI water to fall within this range.

pHFisher scientific accumet portable AP61 pH meterpH units
Dissolved oxygenYSI Model 58 DO Meter 
Nitrates + nitritesHach method 81920.01–0.50 mg L−1 NOmath imageN
NitritesHach method 85070.002–0.30 mg L−1 NOmath imageN
AmmoniaHach method 81550.01–0.50 mg L−1 NHmath imageN
Total nitrogenHach Method 100710.5–25.0 mg L−1 N
Reactive/orthophosphateHach method 80480.02–2.50 mg L−1 POmath image

To investigate the actual values for the reaction rate, the differential rate law equations was used. They were applied to the columns in an attempt to describe the removal in terms of zero, first, and second order kinetics. The concentrations, inverse concentrations, and natural log of concentrations are plotted versus the retention time to show the removals as low order kinetic functions. A linear regression was performed for each condition and the mean squared error was calculated to justify the best fit.


Material Characterization

To determine the difference between natural soil and sorption media in terms of particle size, Figure 2 shows the particle size distribution curves for the Hunter's Trace soil along with media mixes 1 and 2. Hunter's Trace soil is a well graded or evenly distributed soil and the effective size of the soil is 0.16 mm. The effective size is defined as 10% of the sample passing through that sieve size. According to Das [14], effective size is a good indication of intrinsic permeability (i.e., hydraulic conductivity). The selected media mixes are then packed into the other two columns. The particle size distributions and other characteristics for the two media mixes are shown in the following figures. Both the media mixes 1 and 2 are seen to be poorly graded with effective sizes ∼0.08 mm as shown in Figure 2.

Figure 2.

Particle size distributions of soil and media mixes [13].

The material characteristics of the natural Hunter's Trace soil and two sorption media amendments are shown in Table 2. When comparing the media mixes against the natural soil, media mixes have lower density, larger void ratio, larger porosity, smaller specific gravity, and higher permeability. Overall, the performance of the two media mixes appears to be similar. Both media mixes have specific gravities different to the control case (Hunter's Trace Soil) due to the increased amount of organic material. The porosity of the two media mixes is greater than that of the Hunter's Trace soil, thereby allowing for a more rapid flow and smaller HRT.

Table 2. Material characteristics [13].
 Hunter's trace (dry sample)Hunter's trace (moist sample)Media mix 1Media mix 2
Density (g cm−3)1.561.731.411.44
Void ratio0.670.510.560.62
Specific gravity (Gs)2.622.622.192.33
Surface area (m2 g−1)0.130.24
Permeability (cm h−1)62.484.4711.129.19

Results of Column Tests

The control column at 10°C had very low removal efficiencies for nitrogen species. This may be due to slower microbiological nitrification and denitrification resulting from the lower temperature and absence of an electron donor. However, the control column achieved high removal efficiencies for orthophosphate due to the adsorption/adsorption effect. The high dosage case outperformed the low dosage case. The two media mix columns showed similar responses to testing for both influent concentrations of the low and high dosage at 10°C. Both media mixes outperformed the control for the removal of nitrogen species; conversely, they were outperformed by the control case for orthophosphate removal. The media mixes had moderate removal efficiencies for nitrate and total nitrogen. Both media mixes were able to remove more of total nitrogen in the lower dosage case. Media mix 2 had higher removal efficiencies for nitrate in the lower dose case. Media mix 1 removes approximately the same amount of nitrate regardless of the initial nitrate dose. Media mixes 1 and 2 had higher removals of orthophosphate with higher initial phosphorus concentrations. When considering orthophosphate removal, media mix 2 removed twice as much orthophosphate as media mix 1 for the higher dose case, putting its removal capabilities around the average when compared to the control column. This is most likely due to the inclusion of 10% limestone in media mix 2.

The control column at 23°C showed moderate removal efficiencies for nitrogen species. When passing through the Hunter's Trace soil in the control case, nitrite and ammonia levels increased for both low and high dose concentrations. The control column did achieve moderate removals of nitrate when the initial dosage was low. For the case with the higher initial nitrate concentration the control column removed around 26% of the nitrate by the time it reached the second port. But the nitrate levels increased more than the initial dose by the time it reached the bottom causing the overall removal for the control column to be zero. Nitrite levels increased in the column, and the highest amount of nitrite was found at Port 2. These variations reflect the dynamics between nitrification and denitrification. Approximately 28% of the total nitrogen was removed for both dosing situations. The control column achieved higher removal efficiencies for orthophosphate as compared to the situations in these two media mix columns. Again, the high dosage case outperformed the low dosage case. At this temperature, dosing conditions had a greater impact on orthophosphate removals than that at the lower temperature comparatively. The final difference was 30% between the two initial dosing conditions in terms of the removal efficiencies. Besides, the two columns containing media mixes performed similarly with respect to the removal of nitrogen species and orthophosphate. The actual removal efficiencies of these columns improved around 15% for the removal of nitrate and 30% for total nitrogen as compared to the cases at 10°C. In terms of orthophosphate removal, media mix 1 had ∼63% removal efficiency for both dosages, while media mix 2 had 53% for lower dosage and 82% for higher dosages. In this respect, the two media mixes performed as well or better than the control case with lower dosages. The control column does outperform the media mixes overall with a 95% removal efficiency at high doses of orthophosphate.

When the temperature was raised to 28°C, the control column performed quite differently than at the other two lower temperatures. The control column reached its highest nitrate removal at low dosage at Port 2 (77%) instead of the bottom of the column (48%). This is the phenomena of leaching due to the over-consumption of adsorption/absorption capacity. The nitrate removal for high dosage was reached at the bottom of the control column (44%), but was lower than the removal efficiency for low dosage. The highest total nitrogen removal occurred at Port 2 in the control column, and was higher for the low dosage case. Orthophosphate removal followed the same trend as at the other two temperatures; however, it had higher removal efficiency at Port 2 for higher dosages and higher removal efficiency at the bottom of the control column for lower doses. Besides, the two media mixes columns also followed this new trend of achieving higher removal efficiency at Port 2 instead of at the bottom of the column for some instances. Media mixes 1 and 2 had the highest removal efficiency of nitrate at the second port for the low dosage case. But the highest removal efficiency for the high dose continued to be at the bottom of the column. Media mix 1 had approximately the same removal efficiency of around 95% at the bottom of the column. Media mix 2 had a 40% difference in removal efficiency at the bottom, achieving the highest removal efficiency of 90% at high dosage case. Nitrite addition continued to be a problem for both media mixes, with nitrite levels at their highest at ports 1 and 2. Total nitrogen removals were 73.6 and 85.0% at the bottom of the columns containing media mixes 1 and 2, respectively. The only exception is for media mix 1, which showed a 2% greater removal of total nitrogen at Port 2. Orthophosphate removal efficiencies mostly decreased as the water passed through both of the media columns. However media mix 1 has its highest removal efficiency at Port 1 for the low dosage concentrations. This difference, however, was within 1% of the removal efficiency at the bottom of this media column.

Figures 3–5 summarize the nitrate removal through the columns at each temperature. The values for removal efficiency at each port were calculated by averaging the removal efficiency collected at each port combining high and low dosage cases. In the case of 10°C the removal efficiencies seem to follow an increasing trend as the water travels through the column. Media mix 1 presents the highest removal efficiency at each port. Nitrate removals at 23°C appear to follow a similar pattern. However, the columns responded unusually at Port 3 and the removal efficiency of the control case and media mix 1 decreased slightly. In fact, the decrease for media mix 1 of only 0.3% could be considered negligible. Lastly, at 28°C the removal efficiencies increased as the water flowed down the column and media mix 1 always achieved the highest removals. The removal efficiency in the control column decreased slightly between the second and third ports.

Figure 3.

Nitrate removal through columns at 10°C.

Figure 4.

Nitrate removal through columns at 23°C.

Figure 5.

Nitrate removal through columns at 28°C.

Orthophosphate removal through the columns at each temperature is shown in Figures 6–8. Once again, the values for removal efficiency at each port were calculated by averaging the removals at each port combining high and low dosage cases. The data does not appear to follow the same trends as those in nitrate removals with respect to port and temperature. At the lowest temperature, the removal efficiency increased as the water passed through the columns. The control column achieved the highest removal efficiency for most cases. Media mix 1 operated in a similar manner to media mix 2 having just slightly higher removals. At 23°C, all of the columns showed different reactions as the water flowed downward. The control column was outperformed by media mix 2 at the second port, and highest removal of each column was achieved at the bottom port. Media mixes 1 and 2 have similar removals at Port 1 and 3, but media mix 2 has the highest removal efficiency of all the columns at Port 2. Lastly, the 28°C case was considered. The control column achieved the highest removal of all three ports by Port 2 and then removal efficiency was decreased at Port 3. The average removal efficiency of media mix 1 was around 56%. For media mix 2, the removal efficiency amplifies with each port, causing it to go from last place to first.

Figure 6.

Orthophosphate removal through columns at 10°C.

Figure 7.

Orthophosphate removal through columns at 23°C.

Figure 8.

Orthophosphate removal through columns at 28°C.

A comparison of the nitrate removal for the columns at a specific temperature was carried out utilizing an average of the removals at the bottom of the columns for all dosages as shown in Table 3. From this table, it is easy to distinguish how the media mixtures reacted to the change in temperature in terms of nitrate removal. All of the columns achieved their highest removal efficiency at 28°C. Obviously, nitrate removal increased with temperature. Media mixes 1 and 2 had similar increases; with ∼10% increase from 10 to 23°C and then a 15% increase from 23 to 28°C. The control column experienced a 12% increase for the first temperature gap and 25% increase between the higher temperatures. Media mix 1 has the best removal efficiencies for 23, 25, and 28°C with 69.7, 79.7, and 95.3%, respectively. It had the highest nitrate removal for all of the experiments with 95.3% nitrate removal. Overall, media mix 2 is the best on average, and both improvements could be derived from having a better microbiological effect of denitrification.

Table 3. Nitrate removal comparison between temperature and column (final port comparison)
Control case8.00%20.90%45.80%
Media mix 169.70%79.70%95.30%
Media mix 263.20%77.90%93.60%

A comparison of the orthophosphate removal for the columns at a specific temperature was also performed utilizing an average of the removals at the bottom of the columns for all doses Table 4. From this table it is easy to distinguish how the media mixtures reacted to the change in temperature in terms of orthophosphate removal. All columns achieved very high orthophosphate removal at different temperatures; namely the control, media mix 1, and media mix 2 at 10, 23, and 28°C, respectively. However, the difference between the highest and second highest removals for the control was only 1%. The highest overall orthophosphate removal was achieved by media mix 2 at 28°C. The control column outperformed the media mixes for 10 and 23°C and was a fairly close second for 28°C. Overall, media mix 2 and control are the best.

Table 4. Orthophosphate removal comparison between temperature and column (final port comparison)
Control case84.80%80.00%72.10%
Media mix 138.50%63.30%59.90%
Media mix 258.80%67.40%85.50%

Two additional parameters were tested along with the nutrients for each column. The pH levels through the column were taken for use in consideration of real life application of the media. The pH of the stormwater should not be significantly increased or decreased. pH levels around 7 are preferable. The DO levels were also read at each port as the water traveled through the columns. The DO is of special importance when considering the denitrification process. In order for denitrification to occur, the DO levels should be less than 1.0 mg L−1. This information provides insight as to whether the nitrate removals achieved throughout the column could be the result of denitrification. The average pH and DO values can be found in Table 5.

Table 5. Average pH and dissolved oxygen levels in the columns
PortControlMedia mix 1Media mix 2
  1. Note: Negative values indicate increases in pH.


The control column experienced a slight drop in pH for all of the temperatures tested, except for the 23°C. The DO in the control column decreased as the water traveled downward. The DO levels of less than or equal to 1.0 were achieved for the 23°C and 28 cases. However, the DO levels reported for the 10°C case are most likely higher than actual levels in the column due to the reduced flow and increased exposure to the atmosphere during the 10°C scenario. Media mix 1 achieved similar results to the control in terms of pH reduction through the column. The dissolved oxygen levels in the media mix 1 column were almost always around the desired level. The media mix 2 column was the only one able to maintain the pH level for 10°C, increase the pH for 23°C and very slightly decrease the pH for 28°C case. The DO levels reported are similar to the control column.

To discern what effect temperature, dose, and media mixture have on the removal efficiency in our study, the two-way ANOVA analysis was conducted for each dosage case. The removal efficiencies measured at the bottom of each column were used to standardize the data for final comparison. With the existing module in SAS®, a two-way ANOVA table was executed in which removal efficiencies and media were designated as the main effects. The assumptions associated with performance of an ANOVA test, namely independence, normal distributions, and equal variance, were considered across all cases. The assumed alpha value for all of the tests was 0.1, providing 90% level of confidence.

There is a considerable difference between column 1 (media mix 1) and 2 (control), as well as column 2 (control) and 3 (media mix 2). Thus, there is a noteworthy difference between the media mix columns and the control case, but not between each other. However there was sufficient evidence to prove that there is a difference between the temperatures for the low dosage nitrate case. However, there is a considerable difference between the media mix columns and the control, but not between each other. Sufficient evidence is available that there is a difference between 10 vs. 28°C and 23 vs. 28°C for the high dosage nitrate case. Statistically, there is insufficient evidence to suggest a difference in the removal efficiencies between 10 and 23°C. Besides, there is no significant evidence indicating an interaction between column and temperature.

On the other hand, a similar two-way ANOVA table was constructed for orthophosphate in the low dosage case. For all of the effects tested during the orthophosphate low dosage case there is no significant evidence that a difference among column, temperature or interaction between column and temperature exists. The final Two-Way ANOVA table was constructed for Orthophosphate at high dosage. Under 90% level of confidence, there is a considerable difference between column 1 (media mix 1) and 2 (control), as well as column 2 (control) and 3 (media mix 2). There was insufficient evidence to unequivocally show that there is interaction between column and temperature.

Filtration Kinetics

The temperature conversion factor (θ) for each media mix with respect to each nutrient was calculated using the following methodology. The nitrate and orthophosphate removals for each column were plotted separately. Because the initial dose concentrations were not exactly the same, the data from each and every run needed to be plotted individually. Next, a linear regression was executed to deduce the reaction rate of best fit. Tables 6 and 7 summarize the selected rate constants at different temperature.

Table 6. Average nitrate kinetic rate constants
Control (zero order)0.0140.0070.03
Media mix 1 (1st order)0.0120.0170.05
Media mix 2 (zero order)0.0470.0760.07
Table 7. Average orthophosphate kinetic rate constants
Control (zero order)
Media mix 1 (2nd order)0.0010.020.004
Media mix 2 (1st order)

The specific temperature conversion factors (θ) for the media mixes can be calculated by inserting the values for the rate constants and their corresponding temperatures. The temperature conversion factor value was calculated with all three possible permutations; an average of these then produces the value—θavg (see Tables 8 and 9). The permutations in the table below are found by taking the kinetic constants above. The values of 1, 2, and 3 stands for the temperatures 10, 23, and 28°C, respectively, along with the kinetic values associated with those temperatures. For example, θ12 was found using the k values for 10 and 23°C.

Table 8. Temperature conversion constants (θ)
NOmath image[BOND]NControlMedia mix 1Media mix 2
Table 9. Temperature Conversion Constant (θ)
OP (POmath image[BOND]P)ControlMedia mix 1Media mix 2


Based on our complete set of results, the best media for both nitrate and orthophosphate removal was the media mix 2. Yet the best column for nitrate removal for all temperatures is media mix 1. Selection of the best media in terms of orthophosphate was less straightforward. The control column had the best average removals at 10 and 23°C. However, in terms of the media performance, media mix 1 outperforms mix 2 for two of the temperatures, although media mix 2 does have the highest individual occurrence of orthophosphate removal efficiency of all the mixes for all of the temperatures.

The appropriate reaction orders for nitrate turned out to be first order for media mix 1, and zero order for the control and media mix 2. The reaction orders in terms of orthophosphate were different than those for nitrate using our media mixes. A zero order model was selected for the control case, which is similar to nitrate. Media mix 1 and 2 were modeled as second and first order, respectively. Although the reaction rates for the media mixes and the control were selected experimentally there was little difference between the models and thus it is most likely that the reactions are variable order. The temperature conversion constants for nitrate were found to be 1.11, 1.1, and 1.01 for media mix 1, the control and media mix 2, respectively. For orthophosphate they were found to be 1.02, 0.99, and 0.95, respectively. The kinetic rates will be helpful in determining the retention time required for the media mixes to achieve the best removals. Using media characterization information and the desired removal percentages, the depth of media to be employed for pilot studies in two stormwater retention/detention ponds at Ocala, Florida was calculated.


The authors appreciate and acknowledge the support of the Southwest Florida Water Management District and the professional advice and guidance of Chris Zajac from the District.