715 20 Sizing of SSF Wetlands The design of subsurface ow (SSF) wetlands may be roughly divided into two categories: sizing calculations and physical specications. Sizing requires characterization of the incom- ing water and regional climate as well as the goals of wetland treatment, as discussed in Chapter 16. This chapter discusses different sizing methodologies for SSF wetlands. Chapter 21 deals with the physical considerations, including the num- ber of cells, layout, liners, bed depth, media size, plants, and water level control. It is recognized that SSF wetlands are not stand-alone treatment devices but rather form part of an overall treatment process. For HSSF wetlands, primary treatment, at a minimum, is required to remove settleable and oating solids prior to the wetland bed. This is typically provided through the use of settling tanks. Some projects may employ a higher level of pretreatment through the use of activated sludge systems, lagoons, or other treatment processes. Vertical ow (VF) wetlands may or may not have such preliminary treatment stages, and biosolids systems will receive material of differ- ent quality depending on the nature of the overall treatment process. This chapter focuses on the sizing of SSF wetland components, recognizing that this will be only one part of the overall treatment process. To date, most SSF wetlands have been sized using pre- scriptive criteria. The most common approach has been to tie the required wetland size to a parameter that presumably denes the inuent conditions. Inuent loadings, popula- tion equivalents, detention time, and number of bedrooms have all been used to create these scaling factors. Prescrip- tive criteria represent a reasonable approach to wetland siz- ing when there is a large body of preexisting performance data for well-dened pollutants (loading charts), and a more detailed understanding of the internal dynamics of the wet- land is not necessary. As a result, prescriptive criteria are used mainly for the design of SSF wetlands that treat domes- tic wastewater. However, for other projects, there is limited information, and prescriptive criteria cannot be developed. Pilot-scale treatability studies may have to be conducted to measure rate constants and temperature effects. First-order modeling is often employed to interpret these results and make predic- tions on the anticipated performance of a full-scale wetland. Also, rst-order modeling can be used to represent internal proles of pollutant reduction within the wetland. This book employs the P-k-C* model for performance-based wetland design. 20.1 PRESCRIPTIVE SIZING CRITERIA Prescriptive criteria have been the most popular method for sizing SSF wetlands because these methods are quick and mathematically simple. This has led to the widespread use (and sometimes misuse) of prescriptive methods. Prescriptive criteria are only reasonable when they are based on the observed performance of large numbers of wet- lands built to the same physical specications treating the same type of wastewater under the same climatic conditions. If treatment is acceptable under a specied condition, and the data set is large enough to capture inter- and intra-system variability, it may reasonably be expected that future wet- lands operated under the same conditions will also provide acceptable treatment (and that deviations from the mean are acceptably small or infrequent). Unfortunately, prescriptive criteria have been adopted in many situations where there is insufcient data to support their use. LOADING CHARTS Loading charts are design tools, which can be used to size wetland systems. An inuent loading rate is selected to pro- duce a targeted efuent concentration. The wetland area is calculated from the inuent mass load. Recent examples include U.S. EPA (2000a) and Wallace and Knight (2006). Loading charts that can be used in this manner are also included in this book. They are composed, mainly, of wet- land data from Europe, North America, the United Kingdom, and New Zealand; so the vast majority of the systems repre- sented in these charts is from cool temperate climates. For HSSF wetlands, these charts include TSS: Figure 7.30 (C out versus Load in ) BOD: Figure 8.26 (C out versus Load in ) Organic N: Figure 9.18 (C out versus Load in ) Ammonia N: Figure 9.38 (C out versus Load in ) TKN: Figure 9.24 (C out versus Load in ) Total N: Figure 9.30 (C out versus Load in ) Total P: Figure 10.36 (C out versus Load in ) Fecal coliforms: Figure 12.12 (C out versus C in ) For VF wetlands, these charts include TSS: Figure 7.32 (C out versus Load in ) BOD: Figure 8.31 (C out versus Load in ) TKN: Figure 9.25 (C out versus Load in ) © 2009 by Taylor & Francis Group, LLC 716 Treatment Wetlands Because there is considerable scatter in the loading of chart data, selection of the wetland area will likely be an iterative process; selection of a lower inuent loading (larger wetland area) will result in lower estimated efuent concentrations. For instance, consider the case of ammonia in Figure 20.1 (replotted from Figure 9.39). A TKN loading rate of 900 g/m 2 ·yr (approximately 5 m 2 /PE; C i approximately 120 mg/L) results in a predicted efuent ammonia concentration of approximately 15 mg/L. Because of scatter in the data, this estimate of efuent con- centration can only be considered a central tendency. In other words, approximately half of the wetlands loaded at this rate will return ammonia concentrations greater than 15 mg/L. If the loading rate is decreased to 300 g/m 2 ·yr (approxi- mately 15 m 2 /PE) for the same inuent concentration, the probability that the wetland will produce an efuent ammo- nia concentration equal to or less than 15 mg/L is increased considerably. This is consistent with the operational experi- ences with HSSF wetlands, where loadings less than 600 g/ m 2 ·yr (10 m 2 /PE) are necessary to return low-ammonia efu- ents (Geller, 1996). It is tempting to estimate the probability that the wet- land will not perform to this standard by counting the num- ber of data points above the targeted performance goal. For instance, there are seven system·years above the targeted performance level (15 mg/L at 300 g/m 2 ·yr) out of 47 system years at or below this loading rate, so the probability that the wetland will meet the median efuent performance goal is roughly 85%. This predictor only describes median per- formance over the data averaging period used to construct the loading chart. Because these data averaging periods are often long (annual or period of record), important aspects of system performance, such as seasonal or stochastic varia- tions in efuent concentrations, are lost. More importantly, this “point counting” approach does not take into account the effect of the inuent concentration ranges. For instance, many data points in Figure 20.1 are for C i less than 20 mg/L, which is considerably lower than the 120 mg/L inuent con- centration used in this example. Because the data averaging period for Figure 20.1 is annual (198 system·years of data are represented), no infor- mation on seasonal or monthly variation is contained in the chart. This is a problem when the regulatory compliance interval is short (weeks or months). Some design references have attempted to address this by using shorter data-averaging periods, such as system·months (Wallace and Knight, 2006). However, tools such as those shown in Table 9.35, Chap- ter 9, can be used to assist the designer in this regard. As indi- cated in the table, the 90th percentile trend multiplier on the efuent concentrations (1 9) is 1.76. Figure 20.1 indicates that at an inuent loading of 300 g/m 2 ·yr, the central tendency in the efuent ammonia concentrations is approximately 5 mg/L. If the wetland performs at or near this central ten- dency, 90% of the efuent results should be less than 8.8 mg/L (1.76 r 5 mg/L). The probability that the wetland will have poor annual and stochastic results can be crudely estimated by multiplying the probability factors. It should be noted that the amount of data available on the performance of treatment wetlands is much less than stat- isticians are used to working with; thus the application of statistical tools to wetland performance is inherently limited by the lack of current performance data. % ' $ # ! " & FIGURE 20.1 TKN–Ammonia loading chart, with forecasted treatment performance. © 2009 by Taylor & Francis Group, LLC Sizing of SSF Wetlands 717 Tools such as those shown in Figure 9.50, Chapter 9, indicate that some of this variation is likely to be seasonal. There are obvious limitations to the loading chart method. First of all, there must be a functional relationship between the inlet loading and efuent concentration. Figure 7.29 in Chapter 7 is an example where the efuent TSS concentra- tion is essentially independent of the inuent loading. This is because removal of the inuent TSS occurs in only a small region at the inlet of the HSSF wetland bed, and the overall wetland size (and hence, loading) does not play a signicant role in determining efuent TSS concentrations. Secondly, there is considerable scatter in loading chart data, because wetland design, climatic, and loading parameters are not implemented in the same, exact fashion for all systems com- prising the data set. The resulting data scatter means that a large component of the wetland design must necessarily rely on the best professional judgment of the designer. Some per- formance features, such as the role of inuent concentration (independent of inuent load), are also lost. Because loading charts require few mathematical cal- culations, they are an attractive design tool. They have also been misused by designers who do not understand the inher- ent limitations of the loading chart method. It is important to remember that loading charts will not predict system perfor- mance if the physical conguration of the wetland is changed from the parent data set, the pollutant type or concentration changes, or the wetland is in a different climatic region. For instance, loading charts based on the BOD in domestic waste- water will not necessarily apply to industrial efuents where the organic compounds making up the BOD are not the same, and they will likely have different degradation rates (k-rates). Also, a loading chart based on a parameter such as BOD will not predict efuent performance for another pollutant, such as ammonia. Also, as loading charts are based on inlet–outlet data, predictions about pollutant reductions within the wet- land (internal proles) cannot be made. SCALING FACTORS Many efforts have been made to reduce the information con- tained in a loading chart down to a single parameter—a pre- scriptive scaling factor. This is done to further simplify the design process. In general, these scaling factors attempt to describe two aspects of system performance: 1. Median system performance at a specied inlet condition. 2. A “margin of excursion containment” that is intended to allow for intersystem (seasonal and stochastic) as well as intrasystem (differences in the performance response of different wetland systems) variabilities. This margin of excursion containment is rarely dened in the treatment wetland eld; the only exception to date is Wallace and Knight (2006). There is often an underlying assumption that bigger is better (larger wetland area for a given inuent condition), and this has often been used as an excuse to not rigorously collect or analyze performance data. As a result, many pre- scriptive criteria are overly conservative and hinder, rather than help, the reliability of the wetland. There are numerous examples of “bad” or outmoded concepts that are imbedded in various “prescriptive rules.” Examples include making the wetland excessively large (exacerbating freezing prob- lems in cold environments and water loss to ET in arid cli- mates), specifying long length:width ratios (maximizing the cross-sectional loading and associated clogging problems) for HSSF wetlands, and specifying deep beds to increase the hydraulic retention time (exacerbating vertical stratication of the ow). ScalingFactorsforHSSFWetlands Scaling factors provide the simplest approach to wetland design; thus it is not surprising that this is the most com- monly used design method for HSSF wetlands. Examples of some existing criteria include 5 m 2 /PE (BOD loading equivalent to 8 g/m 2 ·d) to produce an efuent BOD of less than 30 mg/L for primary-treated domestic wastewater Source: (EC/EWPCA Emergent Hydrophyte Treat- ment Systems Expert Contact Group and Water Research Centre, 1990) 3.1 cm/d Source: (TVA, 1993) Bed sizing q 5 m 2 /PE (BOD loading equivalent to 8 g/m 2 ·d) Source: (ATV, 1998; ÖNORM B 2505, 2005) 6 g/m 2 ·d of inlet BOD loading to produce an efu- ent with less than 30 mg/L BOD Source: (U.S. EPA, 2000a) 8 g/m 2 ·d of inlet BOD loading to produce a median efuent concentration of 30 mg/L; 5 g/m 2 ·d to pro- duce a median efuent concentration less than 25 mg/L BOD Source: (Wallace and Knight, 2006) 10 to 13 days’ nominal hydraulic retention time Source: (Gustafson et al., 2001) 41 L/m 2 ·d; two to three days’ nominal hydraulic retention time Source: (Lesikar, 1999) 28 m 2 per bedroom per day; trenches 30 cm wide and 92 m long per bedroom Source: (Iowa Department of Natural Resources, 2001) Scaling Factors for VF Wetlands Scaling factors have also been widely employed to size inter- mittently loaded VF wetland systems, especially in Europe. Scaling factors that are currently in use are summarized in Ta ble 15.1, in Chapter 15. • • • • • • • • © 2009 by Taylor & Francis Group, LLC 718 Treatment Wetlands Cautions Regarding the Use of Scaling Factors These summaries are by no means a complete list of all the scaling factors existing in the treatment wetland eld. Obvi- ously, these examples bear little functional relationship to each other; all are clearly artifacts of the available data, reg- ulatory situation, and climatic conditions where they were developed. The designer is thus cautioned that extrapolating these types of prescriptive rules beyond their local regula- tory jurisdiction will likely result in inappropriate wetland sizing; also, many of the physical specications contained in these prescriptive rules are obsolete and have not with- stood the test of modern data analysis in the treatment wet- land eld. EMPIRICAL EQUATIONS Empirical equations are another means of predicting system performance. These equations are developed based on a pre- existing data set (similar to those summarized in the loading charts presented in this book). This data set is then analyzed to determine a mathematical relationship that has the highest level of statistical correlation. It is important to note that this method makes no attempt to describe the internal dynamics of the wetland system and is utterly dependent on the input– output data set being analyzed. Recent examples of this method are included in Crites et al. (2006) and Figure 8.27, Chapter 8, of this book. Empirical equations can best be thought of as a sliding- scale “rule of thumb.” All the limitations inherent in the load- ing chart method and scaling factors also apply to empirical equations. As a result, designers are cautioned that applying empirical equations to situations not representative of the parent data set will often result in inappropriate designs. 20.2 PERFORMANCE-BASED WETLAND SIZING In performance-based design, the effects of degradation rate coefcients (k-rates), temperature (Q-factors), and the com- bined effects of internal hydraulics and pollutant weather- ing (PTIS) are considered discretely. The great advantage of this method is that it allows prediction of treatment wetland performance across different hydraulic and temperature regimes. Historically, early HSSF designs were based on rst-order modeling with the assumption of plug ow (EC/EWPCA Emergent Hydrophyte Treatment Systems Expert Contact Group and Water Research Centre, 1990). In the realm of proprietary wetland designs, this method was extended to waste-specic k-rates and pollutant weathering for soil-media HSSF wetlands (Kickuth, 2002), although these results were not reported to the international scientic community. The rst-order approach was largely abandoned in Europe in favor of simple scaling factors because the major- ity of HSSF wetlands were implemented for small ows of domestic wastewater under similar climatic conditions. In North America, implementation of HSSF wetlands was less rapid than in Europe. However, there has been an expanding interest in using HSSF wetlands for applications other than domestic wastewater treatment, resulting in the continued development of sizing models using rst-order kinetics (Kadlec and Knight, 1996). This book includes a performance-based approach to HSSF wetland sizing using the rst-order P-k-C* model. Conceptually, this is a tank-in-series (TIS) model using a relaxed parameter, P, to account for both hydraulic and weathering effects. The P-k-C* model is discussed in detail in Chapter 6. This performance-based approach is in contrast to most of the currently existing wetland manuals, which advocate a prescriptive loading approach for HSSF wetland design. In this book, loading charts are proposed as a means to check the relative conservativeness of a proposed wetland sizing. The P-k-C* model has a number of advantages for HSSF wetland sizing, which include Loading charts cannot predict internal proles of pollutant reduction, but this is easily done through rst-order modeling. Differences in hydraulic efciency or pollutant weathering can be explicitly addressed. This is especially important when designing full-scale wetlands based on pilot system data, which may have different hydraulic properties. The rst-order approach allows the modeling of sequential removals, such as for nitrogen. Effects of a nonzero background concentration, C*, may be explicitly addressed. The procedure outlined here is for a constant inuent ow situation. Different inuent ow rates may need to be consid- ered to deal with daily or seasonal peaking factors. Different seasons may need to be considered so that the design-limiting (“bottleneck”) period of the year is identied and analyzed. The design-limiting period depends on the goals of the project and the climate of the project location. For instance, for proj- ects in very cold climates, operation of the wetland without freezing during the winter months may be a design priority. In hot, arid climates, water loss due to evapotranspiration (ET ) may be a major concern. BASIC APPLICATION OF THE P-K-C* M ODEL TO HSSF WETLANDS The simplest situation is where there are no gains of water from precipitation or losses of water due to ET or inltra- tion. This can be easily described using the P-k-C* model (Figure 17.1, Chapter 17) and Equation 6.57 in Chapter 6, restated here as Equation 20.1: CC CC kPq k P PP ¤ ¦ ¥ ³ µ ´ * * (/)( /) i V 1 1 1 1 T (20.1) • • • • © 2009 by Taylor & Francis Group, LLC Sizing of SSF Wetlands 719 where k modified first-order areal rate constant, m/d modified first-order volumetric rat V k ee constant, d apparent number of TIS 1 P An example is detailed in Table 20.1. This hypothetical exam- ple is congured for a small HSSF wetland (100 PE) with a BOD loading of 40 g/PE and a ow rate of 200, L/PE (C i 200 mg/L). A k-rate of 45 m/yr is selected based on the distribu- tion shown in Table 8.7 Chapter 8. The wetland is assumed to hydraulically function as eight tanks in series (see Table 6.2, Chapter 6); a reduced value of P 4 has been selected here to account for weathering of the BOD mixture as it passes through the wetland. The background concentration, C*, has been estimated at 8 mg/L based on Figure 8.27, Chapter 8. Equation 20.1 can be solved directly, or the calculation can be completed one tank at a time using a computer spread- sheet, as shown in Table 20.1. WATER BUDGET EFFECTS If there is a net gain or loss of water from the wetland, efuent concentrations will be different from those predicted by Equa- tion 20.1. For HSSF wetlands operating under hot, arid condi- tions, water loss from ET may be a signicant design concern, and the effects of the water budget should be considered. The annual water budget forms the basis for a rst approach to understanding pollutant reductions and the area required. Under the assumption of a constant water level, the ows from the wetland may be computed from the inuent ow rate and meteorological data from the vicinity of the project site. Calcu- lation procedures have been established in Chapter 2. The inlet hydraulic loading will be increased by rain- fall (y0.5–1.5 m/yr) and decreased by ET (y0.5–1.5 m/yr). However, there may be seasonal imbalances. These amounts are important if the wetland is to have a very low hydrau- lic loading, or correspondingly, a long detention time. Net evapotranspiration (ET P) has two effects: lengthening of detention time and concentration of dissolved constituents. The inverse is true in net precipitation environments (P ET). The use of an average ow rate compensates for altered detention time but not for dilution or concentration. There- fore, it may be prudent to consider worst-case seasonal con- ditions when calculating water-budget effects on the wetland treatment performance. Pollutant mass balances are conducted on a cells-in-series basis (see Figure 20.2). Results of the overall water mass balance are apportioned to the cells according the chosen number of TIS. For the rst unit in the series (Equation 6.66 in Chapter 6, restated here as Equation 20.2), QQAPETI 11 i () (20.2) where A ET 1 2 area of the first segment (tank), m e vvapotranspiration, m/d infiltration, m/dI P precipitation, m/d inlet flow rate, m / i 3 Q dd outlet flow rate from segment #1, m /d 1 3 Q The additional data input requirements for the water mass balances are 1. Inow (Q i ) 2. Precipitation (P) 3. Evapotranspiration (ET) 4. Inltration (I) 5. Area (A) TABLE 20.1 Estimated Pollutant Reduction Using a First-Order (P-k-C*) Model for Constant Flows (P ET) Input Parameters Calculated Values Flow rate, Q 20 m 3 /d Volume per tank 14.3 m 3 PTIS (system) 4 Area per tank 125 m 3 Area, A 500 m 2 Inuent ow, Q i 20.0 m 3 /d Porosity, E 0.38 Efuent ow, Q o 20.0 m 3 /d Bed depth 0.30 m Average ow, Q avg 20.0 m 3 /d C i 200 mg/L Efuent mass load 551 g/d C* 8 mg/L Nominal HRT 2.85 d k 45 m/yr HRT based on Q avg 2.85 d k 0.123 m/d HRT based on PTIS 2.85 d C alculated Values S ystem In Exit Tank 1 Exit Tank 2 Exit Tank 3 Exit Tank 4 System Out Net ow m 3 /d 20.0 20.0 20.0 20.0 20.0 20.0 HLR, q m/d 0.040 0.16 0.16 0.16 0.16 0.04 Concentration, C mg/L 200.0 116.4 69.2 42.6 27.5 27.5 Nominal HRT days 2.85 0.71 0.71 0.71 0.71 2.85 © 2009 by Taylor & Francis Group, LLC 720 Treatment Wetlands The example of Table 20.1 is continued forward to exam- ine a situation where there is a net loss of water because of ET. Results of the hydraulic changes are summarized in Table 20.2. For ease of comparison, the parameter P has been kept at P 4 (even though a higher value could be justied in the absence of pollutant weathering). As seen in Table 20.2, there is a net change in the ow rate, hydraulic loading, and detention time as water ows through the wetland reaction segments, because in this par- ticular example, approximately 25% of the water ow is lost through the system. Averaging the inlet and outlet ows does not fully account for these effects. POLLUTANT MASS BALANCES If there is a net gain or loss of water within the system, the ow in each reaction compartment is different, and it becomes necessary to calculate contaminant removals on a mass basis, as concentration alone no longer adequately describes system performance. This computation is then repeated for the remaining seg- ments, in each case using the outlet concentrations and ows from the preceding segment. The wetland outlet concentra- tion is that exiting from the nal reaction segment. The mass output is calculated from the efuent ow rate and output concentration. As the ow rate is different for each reaction segment, Equation 20.1 cannot be used to directly calculate the result. Instead, calculations must be carried out sequentially for each reaction segment (tank). This is most easily done using a com- puter spreadsheet. Comparison of Table 20.1 with Table 20.2 illustrates that the efuent concentrations predicted by these methods are almost identical. Water is lost to ET, which concentrates pollutants within the wetland. However, the volume reduc- tion results in lower hydraulic loading rates (longer detention times), so a greater degree of treatment also occurs within the wetland reactor. Use of average ow accounts for only the longer detention and not the evaporative concentration. Reviewing the system performance from a mass load basis reveals a very different picture. In the example of Table 20.1, the assumption is of no water loss, so the efuent mass load is 27.5 g/m 3 r 20 m 3 /d, or 551 g/d. In the example of Table 20.2, the efuent ow has been reduced by 24%, so the efuent mass load is 27.6 g/m 3 r 15.2 m 3 /d, or 420 g/d. Establish Design Basics Constraint Checks and Iterations Parameter Selection and Calculation Secondary Considerations Set influent flow and concentrations Establish target effluent concentrations Set inflow and seepage Determine precipitation, ET, and temperatures Select k-rates for targeted parameters Select P-values for targeted parameters Select C* values for targeted parameters Estimate wetland area Check proposed wetland site against loading charts Assess seasonal impacts of plant biomass cycling Finalize wetland area Assess the regulatory compliance interval, and select appropriate trend multipliers (1 + Ψ) Adjust k-rate if necessary Adjust k-rate if necessary Adjust k-rate if necessary Check proposed mass demands of the system against biogeochemical constraints (e.g., oxygen transfer) Adjust k-rate if necessary FIGURE 20.2 Design process ow-chart for HSSF wetlands. © 2009 by Taylor & Francis Group, LLC Sizing of SSF Wetlands 721 The difference in the predicted efuent mass load can be approximated by the ratio a (P–ET)/q i ,which is the atmospheric augmentation or decit. The fractional error in a rst-order-model prediction of mass load is approximately equal to a, for a −0.5. Thus, if 25% of the inow evapo- rates, the efuent mass load will be overpredicted by about 25% unless ow corrections are made. If the wetland gains water through precipitation, the inverse is true. It should be noted that averaging the ow will not completely address this problem. Predicted efuent mass loads for the ow assumptions presented in Tables 20.1–20.3 are summarized in Table 20.4. INTERCONNECTED POLLUTANTS Some pollutants have sequential degradation processes and thus require a linkage of the mass balances between the parent and daughter processes. A classic example of this is nitrogen: nitrogen species interconvert, thereby linking the mass balances for organic, ammonia, and oxidized nitro- gen. It is sometimes possible to disconnect these species, as, for instance, in the case of wetlands that receive nitrate but little or no organic or ammonia nitrogen. However, in many cases it is necessary to account for the (internal) production of ammonia from organic sources—the incoming water or the decomposition of wetland necromass—and the internal production of oxidized nitrogen. The reaction sequence has been presented in Equation 17.8, Chapter 17 (restated here as Equation 20.3). A simple presumed chemistry is ORG N NH N NO N N 4x2 l l l (20.3) In a simplied version of this analysis, uptake and return from biomass is not included. The effects of biogeochemical cycling will be explored in a latter part of the design pro- cess. In the case of nitrogen, the three mass balances become TABLE 20.2 Example of BOD Reduction under Variable Flows (ET P) Input Parameters Calculated Values Flow rate, Q 20 m 3 /d Volume per tank 14.25 m 2 Precipitation, P 0.5 mm/d Area per tank 125 m 2 ET 10 mm/d Inuent ow, Q i 20.00 m 3 /d Inltration 0.02 mm/d Efuent ow, Q o 15.24 m 3 /d PTIS (system) 4 Average ow, Q avg 17.62 m 3 /d Area, A 500 m 2 Efuent mass load 420 g/d Porosity, E 0.38 Nominal HLR 0.0400 m/d Bed depth 0.30 m HLR based on Q avg 0.0352 m/d C i 200 mg/L HLR based on PTIS 0.0341 m/d C* 8 mg/L Nominal HRT 2.85 d k 45 m/yr HRT based on Q avg 3.23 d k 0.123 m/d HRT based on PTIS 3.74 d C alculated Values System In Exit Tank 1 Exit Tank 2 Exit Tank 3 Exit Tank 4 System Out Net ow m 3 /d 20.0 18.8 17.6 16.4 15.2 15.2 Precipitation m 3 /d 0.063 0.063 0.063 0.063 0.063 — ET m 3 /d 1.250 1.250 1.250 1.250 1.250 — Inltration m 3 /d 0.003 0.003 0.003 0.003 0.003 — HLR, q m/d 0.040 0.150 0.141 0.131 0.122 0.034 Nominal HRT days 2.85 0.76 0.81 0.87 0.94 3.74 Concentration, C mg/L 200.0 120.5 72.3 43.9 27.6 27.6 TABLE 20.3 Effluent Pollutant Mass Loads (Based on the P-k-C* Model) under Different Flow Scenarios Method of Calculation Predicted Effluent Flow Rate (m 3 /d) Predicted Effluent Concentration (mg/L) Predicted Effluent Mass Load (g/d) P-k-C*, based on constant ow 20 27.5 551 P-k-C*, based on average ow 17.6 27.6 486 P-k-C*; P-ET corrections based on PTIS 15.2 27.6 420 © 2009 by Taylor & Francis Group, LLC 722 Treatment Wetlands linked, and the tank equations are QC Q C I AC k A C C 11 11 1 1OinO,in OOOO * ()()( ) (20.4) QC Q C I AC k A C C k A11 11 1 1AinA,in AAA O ()( )( ) * AAC C 11 () * OO (20.5) QC Q C I AC k A C C k 11 11 1 1NinN,in NNNN A ()( )( ) * AAC C 11 () AA * (20.6) where C C O A organic N concentration, mg/L ammonia NN concentration, mg/L oxidized N concent N C rration, mg/L organic N rate coefficient, O k m/d ammonia N rate coefficient, m/d o A N k k xxidized N rate coefficient, m/d and the subscript “in” denotes parameters associated with the inuent ow for each nitrogen form. These may be rearranged to solve the outlet concentrations for each tank: C QC kAC QIAkA O in O,in O * O 1 1 11 1 ¤ ¦ ¥ ³ µ ´ (20.7) C QC kAC k C C QIAkA A in A,in A A * OO1 O A 1 1 11 1 () * ¤¤ ¦ ¥ ³ µ ´ (20.8) C QC kAC k C C QIAkA N in N,in N N A A A * N 1 11 11 1 * () ¤¤ ¦ ¥ ³ µ ´ (20.9) Equation 20.7 represents the degradation of organic nitrogen. Equations 20.8 and 20.9 contain extra production terms from ammonication in the ammonia balance and from nitrication in the oxidized nitrogen balance. The three must be solved sequen- tially—20.7, followed by 20.8, followed by 20.9. This can be done through an expanded process of the computer spreadsheet il lustrated in Table 20.3, with one sequence of calculations for organic nitrogen, another sequence of calculations for ammonia, and a nal sequence of calculations for oxidized nitrogen. More than One Parameter Multi-parameter wetland sizing has been previously sum- marized in Tables 17.4 and 17.5, Chapter 17, for FWS sys- tems; the same mathematical process can be applied to SSF wetlands. 20.3 ACCOMPLISHING PERFORMANCE- BASED SIZING FOR HSSF WETLANDS The procedure explored here analyzes the effect of changing the wetland area (or equivalently, detention time) on the fore- casted efuent concentrations of contaminants. It is pre- sumed that the designer will conduct the design calculations via a computer spreadsheet so that design changes may be easily explored as an iterative process. The general performance-based sizing algorithm has been previously discussed in Chapter 15 for FWS wetlands. A modied version, more specic to the typical needs and priorities for SSF wetlands, is presented here: Establish the design basis 1. Set inlet concentrations. 2. Set target efuent concentrations (regulatory lim- its and exceedance factors). TABLE 20.4 BOD Removal Based on Performance-Based Criteria; Constant Flow Assumption Input Parameters Calculated Values Flow rate, Q 20 m 3 /d Volume per tank 18.5 m 3 PTIS (system) 4 Area per tank 162.5 m 3 Area, A 650 m 2 Inuent ow, Q i 20.0 m 3 /d Porosity, E 0.38 Efuent ow, Q o 20.0 m 3 /d Bed depth 0.30 m Average ow, Q avg 20.0 m 3 /d C i 200 mg/L Efuent mass load 399 g/d C* 8 mg/L Nominal HRT 3.71 d k 45 m/yr HRT based on Q avg 3.71 d k 0.123 m/d HRT based on PTIS 3.71 d C alculated Values S ystem In Exit Tank 1 Exit Tank 2 Exit Tank 3 Exit Tank 4 System Out Net ow m 3 /d 20.0 20.0 20.0 20.0 20.0 20.0 HLR, q m/d 0.031 0.12 0.12 0.12 0.12 0.031 Concentration, C mg/L 200.0 103.9 55.9 31.9 20.0 20.0 Nominal HRT days 3.71 0.93 0.93 0.93 0.93 3.71 © 2009 by Taylor & Francis Group, LLC Sizing of SSF Wetlands 723 3. Set inow and seepage (typically zero if synthetic liners are used). 4. Determine precipitation, ET, and temperature for critical seasons and targeted outow rates. Parameter selection and primary sizing calculations 5. Select k-rates for the targeted parameters. 6. Assess the regulatory compliance interval, and select appropriate trend multipliers (1 9). 7. Select the PTIS value. Since P incorporates both the hydraulic efciency (number of TIS) and pol- lutant weathering effects, the selected P-value may not be the same for all targeted parameters. 8. Select C* parameters for the targeted parameters. 9. Adjust the wetland area until design goals are met. 10. Check the proposed wetland size against loading charts (if available) to assess the relative conserva- tiveness of the design for the targeted parameters. 11. Check the proposed mass demands of the system against biogeochemical constraints (e.g., oxygen transfer). 12. Adjust the wetland area until design goals are met. Secondary design considerations 13. Assess the seasonal impact of plant biomass cycling (potentially important if nitrogen is a tar- geted parameter). 14. Modify the wetland area, if necessary, to meet secondary considerations. This is necessarily an iterative process, and the designer may have to iterate at several stages in the overall design process, as illustrated in Figure 20.2. Of necessity, performance-based wetlands require more mathematical effort than scaling rules, which is easily accom- plished using spreadsheets and judicious selection of design parameters. This book provides a variety of resources to assist the designer in using performance-based design tools. These include: k-rate, C*, PTIS BOD Table 8.8 Organic Nitrogen Table 9.12 TKN Tables 9.15, 9.33 Total nitrogen Table 9.19 Ammonia (nitrication) Tables 9.25, 9.33 Nitrate (denitrication) Table 9.39 Pathogens Tables 12.15–12.19 Temperature effects (O-factors) BOD Table 8.9 Organic nitrogen Table 9.13 TKN Table 9.17 Ammonia (nitrication) Table 9.32 Seasonal and stochastic variability (1 9), based on cur- rently available data BOD Tables 8.11, 8.12 Ammonia Table 9.35 Total phosphorus Table 10.16 CONSERVATISM IN DESIGN Part of the inherent challenge of using a general design model (such as the P-k-C* model) is that there are so many vari- ables that can be adjusted by the designer. For instance, the designer can introduce conservatism into the design model at many stages, including: Choosing a k-rate at the lower range of the fre- quency distribution Incorporating efuent multipliers for seasonal variations (see Table 8.13, Chapter 8, for example) Choosing a lower value of P (lower hydraulic ef- ciency and/or greater effect of pollutant weathering) Selecting a high concentration for C* (limiting the level of treatment the wetland can achieve) Choosing a higher Q-factor (more temperature sen- sitivity); this is a conservative design assumption in cold climates A designer who selects the most conservative value for each of these parameters will have an overly conservative design that ignores the many thousands of successful SSF wetland systems operating around the world. The current state-of- the-art design knowledge requires judicious selection of design parameters. The recommended approach in this book is to select median parameters for P, k, C*, and Q, predict the system performance, and adjust using variability tables (such as those presented in Tables 8.13, 9.35, and 10.16). Of course, if the physical conguration of the wetland is different than these parent data sets, then designers must use their best pro- fessional judgment to select appropriate design parameters, because existing performance data is unlikely to represent the altered wetland conguration. Fortunately, loading charts (such as those presented in Figures 7.30, 8.26, 9.18, 9.24, 9.30, 9.38, 10.36, and 12.12) provide a simple means to check the relative conservatism of a wetland design against the body of knowledge accumu- lated from wetland systems already in operation. This iter- ative check against loading chart data is also described in Figure 20.3. Let us consider the example wetland previously described in Tables 20.1 and 20.2; the wetland sizing will be determined through system performance rather than a prescriptive scal- ing factor. In Table 20.1, the HSSF wetland size was set at 5 m 2 /PE, which is a scaling factor commonly used in Europe (ATV, 1998). In this example, we will assume that the inlet ow of 20 m 3 /d and inlet BOD concentration of 200 mg/L remain the same and the wetland must achieve a 90% BOD mass load reduction. We will further assume that the k-rate • • • • • © 2009 by Taylor & Francis Group, LLC 724 Treatment Wetlands (45 m/yr) and PTIS (P 4) parameters remain the same as the previous examples for ease of comparison. MOST BASIC CASE: CONSIDERATION OF CONCENTRATION R EDUCTION ONLY,NO CHANGE IN FLOW If it is assumed that there is no change in ow, then a 90% reduction in mass load corresponds to a 90% reduction in concentration. Equation 20.1 can be used (with some alge- braic manipulation) to directly calculate the area required, or a spreadsheet-based approach, such as the one presented in Table 20.1, can be adopted. The spreadsheet can be manu- ally iterated by the designer, or the iteration process can be automated using functions such as the Solver™ routine in Microsoft Excel™. An example of this spreadsheet-based approach (for the assumption of no change in ow) is shown in Table 20.4. The assumption of C* 8 mg/L has been retained from the previous example. Table 20.4 summarizes the manual iteration of Table 20.1. In just three manual iterations, a solution of 650 m 2 (6.5 m 2 / PE) is arrived at. This is a prediction of the median annual performance of the wetland under a constant-ow situation (P ET). SECOND CASE:POLLUTANT REDUCTIONS UNDER VARIABLE FLOW If precipitation and ET are not equal, there will be differ- ences between the calculations for concentration reduc- tion and mass load reduction. The spreadsheet example of Table 20.2 has been utilized here to iterate a solution of 90% mass load reduction for the high ET season previously con- si dered. Results are summarized in Table 20.5. The spreadsheet of Table 20.5 was manually iterated (three times) to converge on a wetland size of 515 m 2 (5.15 m 2 /PE) to achieve the required mass load reduction on median annual basis. ROLE OF C* IN POLLUTANT REDUCTION The pollutant removal rate, k, and the background concen- tration, C*, are independent parameters of the wetland under consideration. However, they are mathematically linked in the sizing process. To use the analogy of descending in an elevator through a building, k describes how rapidly you drop, and C* determines what oor you arrive at. A high value of k implies a rapid drop from the initial inlet con- centration, and a low level of C* implies that the wetland can remove pollutants to low concentrations (if sized large enough). Let us consider the case of Table 20.5, with the altered assumption that C* is now 20 mg/L, not 8 mg/L. This situation could easily occur if there is a signicant por- tion of BOD that is from nondomestic sources. Olive mill processing wastewater and landll leachates are two exam- ples where a higher level of C* may be justied. The results are summarized in Table 20.6, which indicates that the wetland must be 700 m 2 (7.0 m 2 /PE) to achieve the design goal of 90% mass load reduction on a median annual basis (even though the efuent concentration leaving the wetland &("# %)"+ &)"+ ')"+ $* ! FIGURE 20.3 Comparison of the P-k-C* model against existing performance data based on Table 20.2. © 2009 by Taylor & Francis Group, LLC [...]... systems TABLE 20. 9 Rate Coefficients for Vertical Flow Pulse-Loaded Wetlands Compared to Those for HSSF Systems; the VF Values Are for Plug Flow Parameter BOD5 BOD5 NH4-N TN FC VF k-rate (m/yr) HSSF Input (mg/L) HSSF 20th Percentile k-rate (m/yr) HSSF 80th Percentile k-rate (m/yr) HSSF Table Number HSSF PTIS 20 60 — 10–40 12 20 70–95 30–100 100 200 — — — 20 12 2.7 3.3 43 82 62 29.8 18.1 251 8.7 8.7 9.25 9.19... Table 20. 11 Sizing of SSF Wetlands 733 TABLE 20. 11 Examples of Ammonia-N and Nitrate-N Removal/Buildup in VF Constructed Wetlands Location Dhulikhel, Nepala Wapserveen, Netherlandsb Lemmer, Netherlandsb Contrada Petrosa, Italyc Zeutschbach, Austriad Lappach, Austriad a b c d NH4-N In (mg/L) NH4-N Out (mg/L) 27 61 39 97 76 120 0.08 0.23 1 0.1 12.6 25 NO3-N In (mg/L) 0.4 0.04 0.07 0.1 0.7 2.5 NO3-N Out... model (see Tables 20. 3, 20. 6, 20. 8) can greatly facilitate the design compromise between treatment efficiency and the volume of water available for reuse SUMMARY This chapter has integrated the elements of preceding chapters into a sequential approach to designing and evaluating SSF treatment wetlands Sizing is the first step in design, and loading charts, scaling factors, and the P-k-C* model have all... the per-bed loading when in use TABLE 20. 10 Apparent Rate Coefficients for Ammonia for VF Pulse-Loaded Wetlands Stipulations 1 Period of record averages are used in calculations 2 For k-value calculations, the following P-k-C* parameters are selected: a C* 0 mg/L b P 8 TIS 3 Ranges of variables: HLR (cm/d) Mean Median Max Min Results: N·t Percentile 0.05 0.10 0 .20 0.30 0.40 0.50 0.60 0.70 NH4-N In (mg/L)... based on PTIS 20 m3/d 0.5 mm/d 10 mm/d 0.02 mm/d 4 719 m2 0.38 0.30 m 200 mg/L 8 mg/L 45 m/yr 0.123 m/d 20. 49 m3 179.75 m3 20. 0 m3/d 13.16 m3/d 16.58 m3/d 4,000 g/d 225 g/d 94.4% 0.028 m/d 0.023 m/d 0.022 m/d 4.10 d 4.94 d 6.23 d System In m3/d m3/d m3/d m3/d m/d days mg/L © 200 9 by Taylor & Francis Group, LLC Exit Tank 1 Exit Tank 2 Exit Tank 3 Exit Tank 4 20. 0 0.090 1.798 0.004 0.028 4.10 200 .0 18.3... © 200 9 by Taylor & Francis Group, LLC 732 Treatment Wetlands FIGURE 20. 5 Ammonia load response for pulsed VF wetlands (N Further, pulse loading means that water flows to one of the three beds only a fraction of the time it is in service, meaning that there is an instantaneous, per-bed loading as well However, performance (i.e., lower outlet concentrations) increases as loading decreases (Figure 20. 5)... metals is a desired treatment outcome, there obviously must be enough sulfate (on a mol-to-mol basis) in the Sizing of SSF Wetlands 729 influent waters and also enough organic carbon (often supplementally added to the system as peat or compost) to drive sulfate reduction within the wetland (Frostman, 1993) 20. 4 VF WETLANDS (INTERMITTENTLY LOADED BEDS) Vertical flow (VF) nitrification wetlands are vegetated,... based on PTIS 20 m3/d 0.5 mm/d 10 mm/d 0.02 mm/d 4 700 m2 0.38 0.30 m 200 mg/L 20 mg/L 45 m/yr 0.123 m/d 19.98 m3 175.25 m3 20. 00 m3/d 13.33 m3/d 16.66 m3/d 4,000 g/d 400 g/d 90% 0.029 m/d 0.024 m/d 0.023 m/d 4.00 d 4.80 d 6.00 d System In m3/d m3/d m3/d m3/d m/d days mg/L © 200 9 by Taylor & Francis Group, LLC Exit Tank 1 Exit Tank 2 Exit Tank 3 Exit Tank 4 20. 0 0.088 1.753 0.004 0.029 4.00 200 .0 18.3... Table 15.1, Chapter 15, summarizes various scaling factors that have been proposed in the literature First-Order Models The first-order areal model may be used for the calculation of the wetland area as has been detailed for the FWS and HSSF systems The k-values depend on the water-quality parameter and on the different environmental and operational circumstances The ranges reported in Table 20. 9 were... designer must balance the needs for treatment efficiency against the cost of construction The designer could accept a larger (but less efficient) wetland area or construct a smaller, and more efficient, system, but at the added cost of additional berms and control structures CROSS-CHECKS AGAINST EXISTING PERFORMANCE DATA As illustrated in Tables 20. 4 20. 8, application of the P-k-C* model can yield performance . (g/d) P-k-C*, based on constant ow 20 27.5 551 P-k-C*, based on average ow 17.6 27.6 486 P-k-C*; P-ET corrections based on PTIS 15.2 27.6 420 © 200 9 by Taylor & Francis Group, LLC 722 Treatment. m 3 /d 20. 0 20. 0 20. 0 20. 0 20. 0 20. 0 HLR, q m/d 0.031 0.12 0.12 0.12 0.12 0.031 Concentration, C mg/L 200 .0 103.9 55.9 31.9 20. 0 20. 0 Nominal HRT days 3.71 0.93 0.93 0.93 0.93 3.71 © 200 9 by. ow m 3 /d 20. 0 20. 0 20. 0 20. 0 20. 0 20. 0 HLR, q m/d 0.040 0.16 0.16 0.16 0.16 0.04 Concentration, C mg/L 200 .0 116.4 69.2 42.6 27.5 27.5 Nominal HRT days 2.85 0.71 0.71 0.71 0.71 2.85 © 200 9 by