Hydraulic characterization and removal of metals and nutrients in an aerated horizontal subsurface flow “racetrack” wetland treating oil industry effluent Mohammad-Hosein Mozaffari, Ehsan Shafiepour, Seyed Ahmad Mirbagheri, Gholamreza Rakhshandehroo, Scott Wallace, Alexandros I. Stefanakis This is the peer reviewed version of the article, which is published at https://doi.org/10.1016/j.watres.2021.117220. This version is licensed CC-BY-NC-ND 4.0. Mozaffari, M. H., Shafiepour, E., Mirbagheri, S. A., Rakhshandehroo, G., Wallace, S., & Stefanakis, A. I. (2021). Hydraulic characterization and removal of metals and nutrients in an aerated horizontal subsurface flow “racetrack” wetland treating primary-treated oil industry effluent. Water Research, 200, 117220. https://doi.org/10.1016/j.watres.2021.117220. https://creativecommons.org/licenses/by-nc-nd/4.0/ Made available through Montana State University’s ScholarWorks scholarworks.montana.edu Hydraulic characterization and removal of metals and nutrients in an aerated horizontal subsurface flow “racetrack” wetland treating oil industry effluent GRAPHICAL ABSTRACT Hydraulic characterization and removal of metals and nutrients in an aerated horizontal subsurface flow “racetrack” wetland treating primary-treated oil industry effluent Mohammad-Hosein Mozaffari¹, Ehsan Shafiepour², Seyed Ahmad Mirbagheri², Gholamreza Rakhshandehroo3, Scott Wallace4, Alexandros I. Stefanakis5* ¹Civil Engineering Department, Montana State University, Montana, USA ²Khajeh Nasir Toosi University of Technology, Tehran, Iran 3Department of Civil and Environmental Engineering, Shiraz University, Shiraz, Iran 4Naturally Wallace Consulting LLC, Stillwater, MN 55082, USA 5School of Environmental Engineering, Technical University of Crete, GR73100 Chania, Greece * Corresponding author: astefanakis@enveng.tuc.gr Highlights • Novel ‘racetrack’ constructed wetland design tested for the first time • Treatment of oil refinery effluent rich in heavy metals, phenols and nutrients • Positive role of plants and artificial aeration found, even at low retention time of 1.3 days • High heavy metals removal (up to 98%), complete degradation of phenols and ammonia • Higher k-rates calculated than literature findings for other wetland systems 1 1 Hydraulic characterization and removal of metals and nutrients in an aerated 2 horizontal subsurface flow “racetrack” wetland treating primary-treated oil 3 industry effluent 4 Mohammad-Hosein Mozaffari¹, Ehsan Shafiepour², Seyed Ahmad Mirbagheri², Gholamreza 5 Rakhshandehroo³, Scott Wallace4, Alexandros I. Stefanakis5* 6 ¹Civil Engineering Department, Montana State University, Montana, USA 7 ²Khajeh Nasir Toosi University of Technology, Tehran, Iran, 8 ³Department of Civil and Environmental Engineering, Shiraz University, Shiraz, Iran. 9 4Naturally Wallace Consulting LLC, Stillwater, MN 55082, USA 10 5Laboratory of Environmental Engineering and Management, School of Environmental 11 Engineering, Technical University of Crete, 73100 Chania, Greece 12 * Corresponding author: astefanakis@enveng.tuc.gr 13 14 Highlights 15 • Novel ‘racetrack’ constructed wetland design tested for the first time 16 • Treatment of oil refinery effluent rich in heavy metals, phenols and nutrients 17 • Positive role of plants and artificial aeration found, even at low retention time of 1.3 days 18 • High heavy metals removal (up to 98%), complete degradation of phenols and ammonia 19 • Higher k-rates calculated than literature findings for other wetland systems 20 21 Abstract 2 22 Constructed wetlands (CW) are an attractive technology due to their operational simplicity and 23 low life-cycle cost. It has been applied for refinery effluent treatment but mostly single-stage 24 designs (e.g., vertical or horizontal flow) have been tested. However, to achieve a good treatment 25 efficiency for industrial effluents, different treatment conditions (both aerobic and anaerobic) are 26 needed. This means that hybrid CW systems are typically required with a respectively increased 27 area demand. In addition, a strong aerobic environment that facilitates the formation of iron, 28 manganese, zinc and aluminum precipitates cannot be established with passive wetland systems, 29 while the role of these oxyhydroxide compounds in the further co-precipitation and removal of 30 heavy metals such as copper, nickel, lead, and chromium that can simplify the overall treatment of 31 industrial wastewaters is poorly understood in CW. Therefore, this study tests for the first time an 32 innovative CW design that combines an artificially aerated section with a non-aerated section in a 33 single unit applied for oil refinery wastewater treatment. Four pilot units were tested with different 34 design (i.e., planted/unplanted, aerated/non-aerated) and operational (two different hydraulic 35 loading rates) characteristics to estimate the role of plants and artificial aeration and to identify the 36 optimum design configuration. The pilot units received a primary refinery effluent, i.e., after 37 passing through a dissolved air flotation unit. The first-order removal of heavy metals under 38 aerobic conditions is evaluated, along with the removal of phenols and nutrients. High removal 39 rates for Fe (96-98%), Mn (38-81%), Al (49-73%), and Zn (99-100%) generally as oxyhydroxide 40 precipitates were found, while removal of Cu (61-80%), Ni (70-85%), Pb (96-99%) and Cr (60- 41 92%) under aerobic conditions was also observed, likely through co-precipitation. Complete 42 phenols and ammonia nitrogen removal was also found. The first-order rate coefficient (k) 43 calculated from the collected data demonstrates that the tested CW represents an advanced wetland 44 design reaching higher removal rates at a smaller area demand than the common CW systems. 3 45 46 Keywords: aerated wetland; constructed wetland; refinery effluent; phenols; heavy metals; tracer 47 test 48 49 1. Introduction 50 Polluted water from industrial sites represents a major environmental threat for ecosystems and 51 public health by virtue of historic operational practices, incidents and leakages during storage and 52 transport (Langwaldt and Puhakka, 2000; Stefanakis et al., 2020a). These wastewaters typically 53 have a complex composition with a large variety of pollutants of different nature and properties 54 such as heavy metals, organic matter, nutrients, solids, color, salinity, amid of other inorganic and 55 toxic compounds (Wu et al., 2015; Stefanakis, 2018; Ramirez et al., 2019). Among them, heavy 56 metals are considered one the most problematic. Many industrial wastewaters have varying levels 57 of metals that derive from their operation and require further treatment prior the release to the 58 environment or reuse at the operation sites. In particular, heavy metals are frequently found in 59 effluents from refineries and generally the petrochemical industry (Wu et al., 2015; Mustapha et 60 al., 2018; Stefanakis et al., 2018; Jain et al., 2020). 61 Various technologies and processes have been applied for the treatment of refinery effluents, e.g., 62 membrane bioreactors, activated sludge, anaerobic reactors, ozonation etc. (Stefanakis, 2020a; Jain 63 et al., 2020). The green technology of Constructed wetlands (CW) has also been used for petroleum 64 derived water treatment though the respective applications are limited (Ji et al., 2007; Johnson et 65 al., 2008; Wallace et al., 2011; Stefanakis et al., 2016; Stefanakis, 2020a). The positive aspects of 66 CW include lower investment for infrastructure, significantly lower operation and maintenance 67 cost, ecological character, tolerance to flowrate variations, no use of chemicals for the treatment, 4 68 among others (Kadlec and Wallace, 2009; Stefanakis, 2019). In addition, CW can handle varying 69 contaminant loads and treat many constituents of concern at the same time and more effectively 70 than some physical and chemical processes (Stefanakis, 2016; Gorito et al., 2017). These 71 advantages account for the growing interest and increasing number of CW applications for the 72 treatment of various industrial wastewater sources such as from the chemical and petrochemical 73 industry (Wallace et al., 2011; Stefanakis et al., 2016; Stefanakis, 2020a), the oil and gas 74 exploration (Stefanakis et al., 2018; Jain et al., 2020; Stefanakis, 2020b), the food industry such as 75 dairy farms and olive mills (Tatoulis et al., 2017; Wu et al., 2015), and agro-industries such as 76 glass industry, tanneries, cork processing wastewater (Gholipour et al., 2020; Gomes et al., 2018; 77 Ramirez et al., 2019), among others. 78 The application of CW technology for refinery effluent treatment is not widely reported, although 79 current knowledge indicates that CW can be specifically designed and built for the removal of 80 target-constituents from refinery effluents. A few studies on vertical flow CW (VFCW) and 81 horizontal subsurface flow CW designs (HSFCW) are also available in the international literature 82 (Gillespie et al., 2000; Stefanakis et al., 2016; Mustapha et al., 2018; Jain et al., 2020). HSFCW 83 are inherently oxygen-transfer limited systems, thus, anaerobic digestion (methanogenesis) 84 becomes a salient contributor to the removal of the various constituents (Kadlec and Wallace, 85 2009). On the other hand, artificial aeration in CW is an advanced design that increases the 86 phytoremediation potential of CW for heavy metals removal such as Cu+2 (Xin et al., 2019), while 87 it also enhances the biodegradation rate under an aerobic environment (Del Rio et al., 2006; 88 Stefanakis et al., 2019). 89 However, the vast majority of the published literature studies single CW units that utilize only one 90 wetland design (e.g., vertical or horizontal or surface flow), while the aerated CW has barely been 5 91 tested for refinery effluent treatment. This means that a CW system for refinery effluent typically 92 has a high areal footprint. However, the refinery effluent composition would require a combination 93 of different environments, i.e., aerobic and anaerobic conditions, to achieve effective removal rates 94 for the various pollutants especially for nitrogen and heavy metals. This could potentially be done 95 with the use of hybrid (i.e., multistage) CW systems but this would further increase the land area 96 demand. Therefore, the present study proposes and investigates an advanced CW design as a new 97 innovative solution that combines in the same unit different treatment conditions necessary for the 98 effective removal of pollutants of diverse nature aiming at providing a wetland treatment solution 99 with smaller area demand than the existing CW designs. Therefore, the main research goal of this 100 study is to test for the first time this innovative CW design at pilot scale for the effective treatment 101 of process waters from a refinery. The specific goals is to determine the role of plants, substrate, 102 hydraulic retention time and aeration in treating the refinery effluent with focus on the load 1031 removal of a series of heavy metals and nutrients. 0 3 1041 0 4 105 2. Materials and methods 106 2.1. Study location 107 Four pilot-scale HSFCW units (planted and unplanted) with an innovative design were built 108 outdoors at the Khajeh-Nasir Toosi University (35°45'48.0"N 51°24'33.9"E) with different design 109 characteristics, i.e., planted, unplanted (i.e., sand filters), with and without artificial aeration. 110 Wastewater volume was collected from the Tehran oil refinery effluent. This facility, located 111 southern of Tehran City (35°32'24.7"N 51°25'37.7"E), is the second largest refinery in Iran and is 112 one of the oldest refineries with an oil refining capacity of approximately 110,000 barrels per day, 113 and a water consumption of nearly 505 m³/h (Bidhendi et al., 2010). It is considered as the best 6 114 performing refinery for the sake of Environmental Protection Act in Iran. The current wastewater 115 treatment plant of the refinery consists of the DAF unit, followed by an aeration tank, a 116 sedimentation tank and sand filtration unit, while the final effluent is discharged to agricultural 117 fields or recycled in the industrial process. The effluent of the dissolved air flotation (DAF) unit 118 was used for this study, since the DAF system achieves high recovery rates of the residual oil and 1191 it is preferred to remain in operation. 1 9 1201 2 0 121 2.1. Pilot-scale CW design 122 The base unit of the innovative design (called “Racetrack wetland”) of each CW was fabricated 123 out of clear acrylic plastic with a total flow path length of 12.8 m, mean channel width of 7.65 cm, 124 and an effective depth of 0.2 m. The flow path was wrapped in a spiral configuration (Fig. 1a), 125 such that each pilot cell required only 0.86 m² of area. The inflow point was at the centre of the 126 unit and the water followed a meandric subsurface flow path, with the outflow point at one corner 127 of the unit (Figs. 1 and 2). The structural design of the tested ‘Racetrack Wetland’ (RW) system 128 furnishes two different environments (Fig. 2): a planted section without aeration (total length of 129 planted section 6.3 m) and an unplanted section with aeration (total length of unplanted section 5.7 130 m). The aerated unplanted section constituted 44% of the total wetland area and included irrigation 131 tubes in the center of the flow path with specific drippers that provided 0.6 L/min air flowrate at 132 air pressure of 100 mbar. Aeration in the CW was based on previous systems used in North 133 America and Europe (Wallace, 2002), using a 0.175 kW air blower capable of delivering up to 134 626.8 L/m²/min. The unplanted sections were filled with a graded, washed silica sand (size 0.8 - 135 1.2 mm, porosity 24%). The same media was used in the planted sections, but with the addition of 136 5% clay on the top 2 cm, to allow for the plant establishment in the silica soil. The inlet section at 7 137 the center of the unit had dimensions of 26x26 cm, while the outlet section had a length of 80 cm 138 and width 15.5 cm, while coarser gravel (6-7 mm) was used at the inlet and outlet sections to 139 facilitate the flow distribution (Fig. 3). Cyperus alternifolius plants of approx. 1 week old were 140 harvested from local water courses at the distance of 35 kilometers from Tehran Oil Refinery and 141 then immediately planted in the planted section of the CW design, covering 56% of the total 1421 wetland area. 4 2 1431 4 3 8 1441 4 4 Figure 1. (a) Racetrack wetland configuration with aeration (base unit without substrate or 1451 4 5 146 vegetation), and (b) the unplanted and planted sections, the coarse gravel at the inlet and outlet 147 sections, and the influent and effluent sampling points. 148148 9 149149 150 Figure 2. Schematic representation (top view) of the planted and non-planted sections of the 1511 racetrack wetland unit and the meandric flow path. 5 1 1521 5 2 153 Four similar pilot RW units were built and operated in parallel to examine the effects of plant, 154 substrate and aeration in the wetland performance, distinctly, with the following configuration: 155 one pilot unit only with substrate (RW-S), one pilot unit with substrate and forced bed aeration 156 (RW-SA), one pilot unit with substrate planted with Cyperus alternifolius (RW-SP), and one pilot 1571 unit with substrate planted with Cyperus alternifolius and with forced bed aeration (RW-SPA). 5 7 1581 5 8 10 159 2.2. Pilot-scale CW operation 11 160 Once the pilot RW units were built and planted, a solution of 10 mL of a natural liquid fertilizer 161 (NPK 7-4-7, Iranian plant company) mixed with 100 mL of tap water was added to each RW twice 162 per week to cover the nutrients requirement for the initial growth period of the plants and biofilm 163 (Ramond et al., 2012). The feeding with the NPK solution lasted the first 3 months after planting, 164 in order to assist plant adoption to the new environment and sufficient penetration depth of plant 165 roots into the substrate (Qureshi et al., 2005). After that start-up period, tracer tests were carried 166 out for a period of 2 weeks. Once the tracer tests were completed, the refinery effluent was 167 collected in a large tank that fed the CW units and the feeding of the pilot CWs started. 168 The first 3 weeks of operation with the refinery effluent were considered a necessary period for 169 the system to adopt to the feed water after the start-up period, hence no samples were taken (Kadlec 170 and Wallace, 2009; Ramirez et al., 2019). After that period, another batch of refinery effluent was 171 collected and fed, and the sampling and monitoring started. Two flowrates, i.e., 0.20 and 0.67 172 mL/s, were selected corresponding to the target hydraulic retention times (HRT) of 4 days and 1 173 day, respectively. Tracer tests were used to determine more accurately the actual HRT at the two 174 different flowrates. The low flowrate (0.2 mL/s) was first applied to the four CWs to evaluate their 175 removal performance. Afterwards, the high low flowrate (0.67 ml/s) was applied and the samples 176 taking started again after a week to allow for the adoption of the system to the new flowrate 1771 (Schultze-Nobre et al., 2017; Tatoulis et al., 2017; Gomes et al., 2018). 7 7 1781 7 8 179 2.3. Sampling and analytical methods 180 Samples were taken after each HRT cycle, i.e., every 3.2 days for operation period under the first 181 HRT (HLR of 0.2 mL/s) and every 1.3 days for operation period under the second HRT (HLR of 182 0.67 mL/s) and a total of 18 samplings were carried out per pilot unit (9 per HRT). The total 12 183 duration of the experiment was 6 months. Wastewater samples were collected from the influent 184 and effluent points (Fig. 1b) and analyzed immediately to determine the following parameters: pH, 185 Al (aluminum), As (arsenic), Ce (Cerium), Cr (Chromium), Cu (Copper), Fe (iron), K (potassium), 186 Li (lithium), Mg (Magnesium), Mn (Manganese), Mo (Molybdenum), Ni (Nickel), Pb (lead), Sc 187 (Scandium), Se (Selenium), Si (Silicon), Sn (Tin), Sr (Strontium), Ta (Tantalum), Th (Thorium), 188 V (Vanadium), W (Tungsten), Zn (Zinc), total phenols, ammonia nitrogen (NH4-N), nitrate (NO3- 189 N) and orthophosphate (PO4-P). Heavy metals were measured via inductively coupled plasma– 190 mass spectrometry (ICP–MS) according to Standard Method 1640 EPA (APHA, 1999). Nitrogen 191 species (NH4-N, NO3-N) were measured via spectrophotometry (T80 UV, PG instrument LTD) 192 according to Standard Method 4500-NO -3 N and ASTM D1426 for NH4-N. Orthophosphate (PO4- 193 P) were measured via spectrophotometry (T80 UV, PG instrument LTD) according to Standard 194 Method 4500-P-C. Total phenols was measured via spectrophotometry (T80 UV, PG instrument 195 LTD) using Standard Method ASTM D1783, and pH was determined by a VIVOSUN PH Meter 196 Digital PH Tester Pen for Water. Conductivity was measured with a conductivity meter (MS1- 1971 IMP403-24) and converted to salinity. Tracer tests were conducted using NaCl. 9 7 1981 9 8 199 2.4. Calculations and statistical analysis 200 To estimate the removal performance, the removal rates of the various parameters were calculated, 201 assuming that the rate coefficients can be described by the means of a first-order removal laws. 202 Considering the flowrates, the HRT, the CW surface area and the concentrations of the various 203 parameters, the first-order reaction rate coefficients (k-rates) for wastewater applied and treated in 204 HSFCW were calculated based on the equation (Tarutis et al., 1999; Kadlec and Wallace, 2009): 205 k = (Q x (lnCin – lnCout))/A 13 206 where, Q is the inflow (m³/day), A the surface area (m²), Cin the influent concentration (mg/L), 207 Cout the effluent concentration (mg/L) and k the first-order rate coefficient (m/d). 208 First-order reaction rate coefficients (k-rates) were calculated from the observed removals and 209 tracer studies (number of the tanks in series) and were compared to those previously reported in 210 the literature. The nominal HRT was calculated based on the void volume of the saturated media 211 without considering the inlet and outlet sections of the units. 212 Statistical analysis was performed using the SPSS software (SPSSInc., Chicago, IL, USA; Version 213 12.0). The data was analysed through one-way analysis of variance (ANOVA) at the 95% 214 significance level (p ≤ 0.05) to compare the performance of the CW units in the removal of the 215 various parameters. Post Hoc pair comparisons were also performed to test equal variations, using 216 Tukey’s honestly significant difference test. Homogeneity of variance tests (Levene) were 2172 bypassed, since the number of data points for each group was the same. 1 7 2182 1 8 219 3. Results and discussion 220 3.1. Tracer experiment results 221 Prior to start measuring the treatment performance, tracer experiments were performed in the RW- 222 SP unit (non-aerated, planted system) at the two different flowrates, i.e., 0.67 and 0.2 mL/s that 223 correspond to the target HRT of 1 day and 4 days, respectively. Sodium chloride (NaCl) was used 224 as a tracer at 1,000 mg/L concentration. Electrical conductivity was measured to obtain the 225 response curve. Tracer data was analyzed using the gamma function distribution method (Kadlec 226 and Wallace, 2009). The actual HRT values were 1.37 and 3.16 days, respectively, with the 227 associated number of tanks-in-series (NTIS) of 6.4 and 12.32 (Fig. 3). Tracer recoveries were 70% 228 and 51% for the high and low flowrate, respectively. 14 2292 2 9 Figure 3. Tracer response curve for the targeted 1-day and 4-days HRT. 2302 3 0 2312 3 1 232 Testing the system at the two different flowrates indicated different hydraulic performances. While 233 the variation in the HRT was roughly proportional to the applied flowrate and more or less 234 expected, the degree of internal mixing (NTIS) shifted from 12.32 (HRT of 3.16 days) to 6.4 (HRT 235 of 1.37 days). This was likely caused by higher flow velocities at the higher flowrate, resulting in 236 a greater degree of internal mixing, despite the length to width ratio of the system being 15 237 approximately 7.7. As a point of reference, the reported mean NTIS value for HSFCW over 37 238 tracer studies is 11.0 (Kadlec and Wallace, 2009). It is expected that the operation of the aeration 16 239 system would increase the internal mixing (Boog, 2013; Boog et al., 2014). A comparable aerated 240 HSFCW at an experimental facility in Germany had an NTIS value of 4.5 while continually aerated 2412 (Boog, 2013), although the system had a length to width ratio of 3.9. 4 1 2422 4 2 243 3.1. Overall treatment performance 244 Table 1 shows the influent and effluent values and the respective removal rates of the various 245 parameters at the flowrate of 0.2 mL/s (HRT of 3.16 days) and Table 2 at the flowrate of 0.67 mL/s 246 (HRT of 1.37 days). First-order reaction rate coefficients (k-rates) were calculated based on the 247 observed removals and tracer tests and compared to other results in the literature (Table 3). The 248 removal rates of the various heavy metals are presented in Fig. 4 for both HRT tested, while Fig. 249 4 shows the removal rates for nitrogen, phosphorus and phenols. 250 In general, the aerobic conditions in the wetland (aided by the aeration system) allowed relatively 251 high removal rates for Fe (96-98%), Mn (38-92%), Al (63-73%), and Zn (99%) at both HRT (Fig. 252 3), generally as oxyhydroxide precipitates. Removal of Cu (61-80%), Ni (70-85%), Pb (96-99%) 253 and Cr (63-92%) under aerobic conditions was also observed (Fig. 3), likely through co- 254 precipitation. These removal rates appear to be higher than typically reported figures for passive 255 CW systems (Aslam et al., 2007; Mustapha et al., 2018). The average pH values of the influent 256 (6.8-7.5) and effluent form all units (7.1-7.5), i.e., close to the neutral zone, also indicate that there 257 was no negative interference to the major metal removal mechanisms (i.e., adsorption, 258 precipitation, sedimentation, plant uptake) (Nyquist and Greger, 2009). The effluent pH values 259 were also within the Iranian standards for environmental discharge and effluent reuse (6-8.5; IEAP, 2602 6 261261 0 17 2008). 18 262 Table 1. Average influent (IN) and effluent (EFF) concentrations (μg/L), respective treatment 263 performance (Removal, %) and inlet mass load (g/m²/yr) of the pilot RW units at the HRT of 1.37 264 days (n=9). RW-S RW-SA RW-SP RW-SPA Inlet mass load IN EFF Rem EFF Rem EFF Rem EFF Rem (g/m²/yr) Parameter (µg/L) (µg/L) (%) (µg/L) (%) (µg/L) (%) (µg/L) (%) Al 300 110 63.3 105 65.0 100 66.7 80 73.3 2.20 As 4.37 2.43 44.4 2.15 50.8 2.25 48.5 2.16 50.6 0.032 Ce 1 0.1 90.0 0.26 74.0 0.05 95.0 0.054 94.6 0.007 Cr 6.19 1 83.8 1.52 75.4 0.48 92.2 0.62 90.0 0.045 Cu 30.58 7.76 74.6 11.95 60.9 6.21 79.7 8.0 73.8 0.224 Fe 3580 70 98.0 160 95.5 68 98.1 142 96.0 26.3 K 6510 3440 47.2 4110 36.9 1860 71.4 2080 68.0 47.7 Li 34.62 20.32 41.3 26.61 23.1 21.31 38.4 25.2 27.2 0.254 Mg 27900 22290 20.1 27050 3.0 22170 20.5 22980 17.6 204.6 Mn 130 80 38.5 30 76.9 80 38.5 25 80.8 0.953 Mo 22.29 9.33 58.1 12.83 42.4 4.62 79.3 6.58 70.5 0.163 Ni 52.96 15.95 69.9 14.5 72.6 16.01 69.8 7.87 85.1 0.388 Sc 18.53 4.32 76.7 3.73 79.9 2.9 84.3 1.8 90.3 0.136 Se 14.27 8.97 37.1 8.89 37.7 6.5 54.4 7.1 50.2 0.105 Si 10190 8630 15.3 8040 21.1 6410 37.1 7001 31.3 74.7 Sn 1.57 1.37 12.7 1.49 5.1 0.65 58.6 0.98 37.6 0.012 Sr 2480 1390 44.0 1710 31.0 1480 40.3 2025 18.3 18.2 Ta 0.25 0.13 48.0 0.17 32.0 0.11 56.0 0.16 36.0 0.002 Th 1.05 0.09 91.4 0.1 90.5 0.06 94.3 0.07 93.3 0.008 V 101 3.36 96.7 3.64 96.4 2.63 97.4 2.52 97.5 0.741 W 2.28 0.68 70.2 1.35 40.8 0.39 82.9 0.85 62.7 0.017 Zn 12820 80 99.4 180 98.6 113 99.1 128 99.0 94.0 19 Pb 62 1.78 97.1 1.82 97.1 0.8 98.7 1.1 98.2 0.455 Phenols 638 3.3 99.5 2.9 99.5 1.7 99.7 0.9 99.9 4.679 NO3-N 18000 12100 32.8 13900 22.8 8300 53.9 9100 49.4 132.0 NH4-N 31000 11300 63.5 20 99.9 20 99.9 18 99.9 227.4 PO4-P 20600 20600 0.0 20600 0.0 1000 95.1 1000 95.1 151.1 265265 266266 267 Table 2. Average influent (IN) and effluent (EFF) concentrations (μg/L), respective treatment 268 performance (Removal, %) and inlet mass load (g/m²/yr) of the pilot RW units at the HRT of 3.16 269 days (n=9). RW-S RW-SA RW-SP RW-SPA Inlet mass load IN EFF Rem EFF Rem EFF Rem EFF Rem (g/m²/yr) Parameter (µg/L) (µg/L) (%) (µg/L) (%) (µg/L) (%) (µg/L) (%) Al 300 153 49.0 120 60.0 108 64.0 91 69.7 7.38 As 4.37 3.2 26.8 2.9 33.6 3.1 29.1 0.9 79.4 0.11 Ce 1 0.12 88.0 0.13 87.0 0.05 95.0 0.05 95.0 0.02 Cr 6.19 2.3 62.8 2.5 59.6 0.48 92.2 0.62 90.0 0.15 Cu 30.58 8.8 71.2 13.4 56.2 7.01 77.1 8.8 71.2 0.75 Fe 3580 91 97.5 95 97.3 71 98.0 152 95.8 88 K 6510 2,050 68.5 4,110 36.9 1,930 70.4 760 88.3 160 Li 34.62 13.86 60.0 30.1 13.1 19.5 43.7 21.5 37.9 0.85 Mg 27900 15,390 44.8 20,015 28.3 14,510 48.0 15,310 45.1 686 Mn 130 30 76.9 10 92.3 39 70.0 10 92.3 3.20 Mo 22.29 7.37 66.9 18.4 17.5 5.93 73.4 6.61 70.3 0.55 Ni 52.96 12.18 77.0 11.63 78.0 12.52 76.4 13.71 74.1 1.30 Sc 18.53 4.29 76.8 3.89 79.0 4.2 77.3 3.5 81.1 0.46 Se 14.27 5.4 62.2 5.8 59.4 3.5 75.5 3.55 75.1 0.35 20 Si 10190 8,040 21.1 8,005 21.4 7,011 31.2 7,000 31.3 251 Sn 1.57 0.97 38.2 1.54 1.9 0.73 53.5 1.29 17.8 0.04 Sr 2480 890 64.1 2,130 14.1 980 60.5 1,560 37.1 61 Ta 0.25 0.2 20.0 0.24 4.0 0.06 76.0 0.08 68.0 0.01 Th 1.05 0.09 91.4 0.11 89.5 0.07 93.3 0.08 92.4 0.03 V 101 3.57 96.5 3.5 96.5 3.1 96.9 3.2 96.8 2.48 W 2.28 0.79 65.4 0.79 65.4 0.79 65.4 0.79 65.4 0.06 Zn 12820 60 99.5 130 99.0 78 99.4 91 99.3 315 Pb 62 2.6 95.8 2.62 95.8 1.3 97.9 1.36 97.8 1.52 Phenol 638 7.8 98.8 4.9 99.2 18 97.2 2.3 99.6 15.7 NO3-N 18000 16500 8.3 17300 3.9 1100 93.9 2400 86.7 442.4 NH4-N 31000 17300 44.2 20 99.9 6500 79.0 20 99.9 761.6 PO4-P 20600 20600 0.0 20600 0.0 900 95.6 890 95.7 506.1 270270 21 2712 7 1 Figure 4. Removal rates of heavy metals in the four pilot CW units at the two HRT tested; low 2722 7 2 22 273 (1.37 days) and high (3.16 days). 23 274 Phenols were completely removed in the units under both HRT (Fig. 4), which is in agreement 275 with results from other studies (Stefanakis et al., 2016; Schultze-Nobre et al., 2017; Gomes et al., 276 2018; Jain et al., 2020). NH4-N reached high removal in the planted units (Fig. 4) and was 277 practically completely removed in the aerated RW units apparently due to the higher dissolved 278 oxygen availability provided by plant roots and mostly by artificial aeration for the nitrification 279 process (Kadlec and Wallace, 2009; Mustapha et al., 2018; Stefanakis et al., 2019). As a result, 280 the first-order reaction rate coefficients (k; Table 4) estimated for these parameters were limited 281 by the complete removal of the mass loading, and further investigation of removal of these 282 parameters, especially phenols, in aerated CW is warranted. NO3-N removal was limited at a 283 certain level due to the enhanced aeration and conditions conducive to extensive denitrification 284 did not occur in the tested pilot configurations, although higher removal rates were measured in 285 the planted units probably due to higher plant uptake (Stefanakis et al., 2019). Phosphorus removal 286 in the system was attributable to short-term plant biomass uptake over the duration of the study 2872 period, as indicated by the high removal rates measured only in the planted units. 8 7 2882 8 8 24 2892 8 9 Figure 5. Removal rates of phenols, ammonia nitrogen, nitrate and phosphorus in the four pilot 2902 9 0 291 CW units at the two HRT tested; low (1.37 days) and high (3.16 days). 292292 293 Table 3. First-order reaction rate coefficients (k) for the racetrack wetland design and comparison 294 with available literature data. Parameter This Study Other Studies 219–290 m/yr (1.37 d HRT, range for all units) 42 m/yr (Manyin et al., 1997) Fe 82-107 m/yr (3.16 d HRT, range for 66 m/yr (Tarutis et al., 1999) all units) 169 m/yr (1.37 d HRT with Mn 21 m/yr (SF; Tarutis et al., 1999) aeration) 25 71-88 m/yr (1.37 d HRT without 10-60 m/yr (HSF with river gravel; Sikora et al., aeration) 2000) 100-600 m/yr (HSF with limestone; Sikora et al., 2000) 54-97 m/yr (1.37 d HRT, range for all units) Cu 80 m/yr (SF, estimated; Kanagy et al., 2008) 23-39 m/yr (3.16 d HRT, range for all units) 81-82 m/yr (1.37 d HRT, range for all units) Ni 15-18 m/yr (SF+HSF; Sinicrope et al., 1992) 29-42 m/yr (3.16 d HRT, range for all units) 219-284 m/yr (1.37 d HRT, range for all units) Pb 104 m/yr (SF, estimated; Kanagy et al., 2008) 95-121 m/yr (3.16 d HRT, range for all units) 52-57 m/yr 1.37 d HRT, unplanted) 148-168 m/yr (1.37 d HRT, 25-35 m/yr (SF+HSF; Sinicrope et al., 1992) planted) Cr 20-80 m/yr (SF; Kadlec and Srinivasan, 1995) 34-46 m/yr (3.16 d HRT, 3-64 m/yr (Kadlec and Wallace, 2009) unplanted) 59-66 m/yr (3.16 d HRT, planted) 28 m/yr (SF+HSF; Sinicrope et al., 1992) 360-450 m/yr (1.37 d HRT, range Zn 55-121 m/yr (SF; Gillespie et al., 2000) for all units) 21-42 m/yr (SF, estimated; Dorman et al., 2009) 26 118-146 m/yr (3.16 d HRT, range for all units) 8-17 m/yr (HSF, estimated; Rossmann et al., 2012) 107.2 m/yr (HSF; Stefanakis et al., 2016) 96 -165 m/yr (1.37 d HRT, planted 98.9 m/yr (HSF; Tatoulis et al., 2017) with aeration) Phenols 100 m/yr (SF, estimated; Polprasert and Dan, 1994) 152-201 m/yr (3.16 d HRT, planted 76 m/yr (SF, unplanted; Tang and Lu, 1993) with aeration) 150 m/yr (SF, planted; Tang and Lu, 1993) 100 m/yr (SF, estimated; Dong and Lin, 1994) 960,6 m/yr (VF, planted with aeration; Stefanakis et 233-237 m/yr (planted with al., 2019) aeration) 589 m/yr (HSF, planted with aeration; Stefanakis, 13.8 m/yr (1.37 d HRT, unplanted, 2020b) NH4-N no aeration) 2.9 m/yr (HSF, planted without aeration; Nivala et 24.4 m/yr (3.16 d HRT, unplanted, al., 2019) no aeration) 22.4 m/yr (HSF, planted without aeration; Stefanakis et al., 2016) 75 m/yr (HSF, planted with aeration; Nivala et al., 0.9-2.0 m/yr (1.37 d HRT, 2019) unplanted) 65.1 m/yr (VF, planted with aeration; Stefanakis et 51-73 m/yr (1.37 d HRT, planted) al., 2019) NO3-N 6.0-9.3 m/yr (3.16 d HRT, 96.4 m/yr (HSF, planted with aeration; Stefanakis, unplanted) 2020b) 16-19 m/yr (3.16 d HRT, planted) 3.9 m/yr (HSF, planted without aeration; Nivala et al., 2019) 27 295295 296 3.2. Removal of metals 297 As seen in Tables 1 and 2, a removal rate was detected for all measured metals, albeit at varying 298 mass loading rates and removal percentages. Generally, heavy metals removal can be attributed to 299 the (i) adsorption onto the wetland substrate (sand, gravel) and organic matter (plant roots), (ii) 300 air/water interactions, which generally lead to an aerobic (oxidizing) environment, especially in 301 the pilot units with artificial aeration, resulting in precipitation and co-precipitation of metals, and 302 (iii) uptake and assimilation in plant biomass (Sheoran and Sheoran, 2006; Kadlec and Wallace, 303 2009; García et al., 2010). 304 Iron (Fe) loading rates were 26 and 88 g/m²/yr at the two flowrates with corresponding removals 305 of 96-98% (Fig. 3). These loading rates are far below those recommended for CWs treating mine 306 drainage, which are in the range of 3,650 – 7,300 g/m²/yr (Hedin et al., 1994; PIRAMID 307 Consortium, 2003; Nyquist and Greger, 2009). The Fe effluent values of all pilot CWs were also 308 below the national limit for environmental discharge (300 μg/L; IEAP, 2008). The absence of high 309 organic loads and the respectively limited oxygen demand within the wetlands in the present study, 310 as well as the external aeration provided to the system, generally supports the hypothesis that Fe 311 was removed as an oxyhydroxide precipitate (Fe(OH)3). It is proposed that Fe removal in CWs is 312 a first-order process, with reaction rate coefficients (k) estimated at 42 m/yr (Manyin et al., 1997) 313 and 66 m/yr (Tarutis et al., 1999). Analysis of the data from this study yielded k-rates between 82- 314 107 m/yr at the longer HRT (3.16 days) and 219-290 m/yr at the shorter retention time (1.37 days) 315 (Table 3). Since virtually all Fe was removed in the CW, the higher reaction rate coefficients 316 indicate that Fe removal occurred quickly within the system. 28 317 Manganese (Mn) loads were 0.95 and 3.2 g/m²/yr at the two flowrates (Tables 1 and 2), which is 318 considerably lower than the 182.5 g/m²/yr recommended for CWs treating mine drainage 319 (PIRAMID Consortium, 2003). Mn removal was consistently higher in the CW units with artificial 320 aeration at both flowrates. At the HRT of 1.37 days, Mn removal was 70-77% without aeration 321 and 92% with aeration (Fig. 3); for the HRT of 3.16 days, Mn removal was 38.5% without aeration 322 and 77-81% with aeration (Fig. 3). As the kinetics for Mn oxidation and precipitation are generally 323 slower compared to Fe (Hedin and Nairn, 1993) and Mn removal is not significant in aerobic CWs 324 with low Fe concentration (Hallberg and Johnson, 2005), the enhancement of the aerobic 325 conditions in the aerated wetlands could potentially have aided in Mn removal. The Mn effluent 326 values of all pilot CWs were also below the national limit for environmental discharge (50 μg/L; 327 IEAP, 2008). Mn removal as a first-order process in CWs has been proposed. Data analysis 328 indicated reaction rate coefficients (k) of 169 m/yr for the optimum conditions (highest loading 329 with aeration), and 71-88 m/yr (highest loading without aeration) (Table 3). In the literature, 21 330 m/yr has been proposed for surface flow (SF) CW (Tarutis et al., 1999). For HSFCWs (without 331 aeration), rates of 10-60 m/yr for river gravel and 100-600 m/yr for limestone-based systems have 332 been estimated (Sikora et al., 2000). 333 Aluminum (Al) can form hydroxide precipitates under oxidizing conditions, which are strongly 334 associated with phosphorus retention. Al loads were 2.2 and 7.4 g/m²/yr (Tables 1 and 2), 335 compared to loads of approximately 0.44 g/m²/yr observed for municipal wastewater applications 336 of non-aerated HSFCWs (Vymazal and Krása, 2003). The removal percentages in this study 337 ranged from 63-73% at the HRT of 3.16 days, and 49-70% at the HRT of 1.37 days (Fig. 3), 338 indicating that longer HRT resulted in better Al removal. Respective figures for Al reported in the 339 literature are approximately 70% (Gensemer and Playle, 1999), and are comparable to this study. 29 340 Copper (Cu) loads were 0.75 and 0.22 g/m²/yr at the two flowrates (Tables 1 and 2), with removal 341 percentages between 56-80%. While Cu removal is often associated with reducing conditions and 342 sulfide precipitation, removal under aerobic conditions via co-precipitation with Fe and Mn 343 oxyhydroxide precipitates is also possible. For mining applications, loading rates of 3,650 g/m²/yr 344 (sulfide producing conditions) and 18.25 g/m²/yr (aerobic conditions) have been proposed 345 (PIRAMID Consortium, 2003), which are considerably higher than this study. The Cu effluent 346 values of all pilot CWs were below the national limit for environmental discharge (20 μg/L; IEAP, 347 2008). Cu removal as a first-order process under aerobic conditions is a possibility (Tarutis et al., 348 1999). If Cu removal is through co-precipitation with Fe and Mn, and those metals follow a first- 349 order process, reaction rate coefficients (k) can also be estimated for Cu. Data analysis indicates 350 k-rates of 23-39 m/yr at the lower HRT, and 54-97 m/yr at the higher HRT (Table 3). In a pilot- 351 scale SFCW treating oil produced water, Cu removal data indicated a k-rate of 80 m/yr (Kanagy 352 et al., 2008). 353 Nickel (Ni) loads were 1.30 and 0.39 g/m²/d at the two flowrates, with removal percentages 354 between 70-85% (HRT of 3.16 days) and 74-77% (HRT of 1.37 days) (Tables 1 and 2). While Ni 355 removal is often associated with reducing conditions and sulfide precipitation, removal under 356 aerobic conditions through co-precipitation with Fe and Mn oxyhydroxide precipitates is also 357 possible. For mining applications, loading rates of 730 g/m²/yr (sulfide producing conditions) and 358 14.6 g/m²/yr (aerobic conditions) have been proposed (PIRAMID Consortium, 2003), which are 359 considerably higher than this study. Ni removal as a first-order process under aerobic conditions 360 is a possibility (Tarutis et al., 1999), but data sets to estimate a first-order reaction rate coefficient 361 (k) are limited. Data from a SF + HSF CW at Sacramento, California resulted in estimated k-rates 362 of 15-18 m/yr (Sinicrope et al., 1992), at Ni loading rates of 5.5-6.7 g/m²/yr (Kadlec and Wallace, 30 363 2009). Data from this study yields k-rate estimates of 29-48 m/yr (HRT of 3.16 days) and 81-82 364 m/yr (HRT of 1.37 days) (Table 3). 365 Lead (Pb) loads were 1.52 and 0.46 g/m²/yr at the two flowrates (Tables 1 and 2). Almost all Pb 366 was removed (96-99%; Fig. 3), which is consistent with the formation of very insoluble sulfate 367 and carbonate precipitates (DeVolder et al., 2003). The Pb effluent values of all pilot CWs were 368 below the national limit (50 μg/L; IEAP, 2008). Pb removal as a first-order process has been 369 proposed, but previous data sets did not allow for the estimation of a k-rate (Kadlec and Wallace, 370 2009). Data from this study yields k-rate estimates of 95-121 m/yr at the lower HRT and 219-284 371 m/yr at the higher HRT (Table 3). Since virtually all Pb was removed at both HRTs, the use of the 372 higher k-rate estimates appears justified. In a pilot-scale SFCW treating oil produced water, Pb 373 removal data indicated a k-rate of approximately 104 m/yr (Kanagy et al., 2008). 374 Chromium (Cr) loads were 0.15 and 0.05 g/m²/yr at the two flowrates (Tables 1 and 2). Cr removal 375 was affected by plants’ presence. The removal in the unplanted units were 75-84% (low loading 376 rate) and 60-63% (high loading rate), and 90-92% in the planted units at both loading rates (Fig. 377 3). On the other hand, the artificial aeration did not appear to influence Cr removal. The Cr influent 378 and effluent values of all pilot CWs were below the national limit for environmental discharge (50 379 μg/L; IEAP, 2008). Cr can be removed through co-precipitation with Fe and Mn oxyhydroxide 380 precipitates; the first-order modeling approach applied to those metals can also be applied to Cr. 381 A summary of Cr removal data estimated k-rate coefficients ranging from 3-64 m/yr, with a median 382 value of 19 m/yr (Kadlec and Wallace, 2009). Data from SFCW and HSFCW mesocosms at 383 Sacramento, California treating municipal wastewater yielded k-rate estimates of 25-35 m/yr 384 (Sinicrope et al., 1992), and data from a SFCW mesocosm treating petroleum wastewater yielded 385 k-rate estimates of 20-80 m/yr (Kadlec and Srinivasan, 1995). In a pilot-scale SFCW treating ash 31 386 basin water, Cr removal data suggested k-rates of 29-36 m/yr (Dorman et al., 2009). In this study, 387 the unplanted units resulted in k-rate estimates of 34-46 m/yr (HRT of 3.16 days) and 52-57 m/yr 388 (HRT of 1.37 days) and the planted units in 59-66 m/yr (HRT of 3.16 days) and 148-168 m/yr 389 (HRT of 1.37 days) (Table 3). 390 Zinc (Zn) loads were 315 and 94 g/m²/yr at the two flowrates (Tables 2 and 3). Zn was almost 391 completely removed (99%; Fig. 3), which is consistent with the formation of insoluble carbonate 392 precipitates, and through co-precipitation with Fe and Mn oxyhydroxide precipitates (Younger et 393 al., 2002; Kadlec and Wallace, 2009). In comparison, recommendations for loading rates for mine 394 water treatment range from 14.6 g/m²/yr (PIRAMID Consortium, 2003) up to 2,555 g/m²/yr 395 (Younger et al., 2002). The Zn effluent values of all pilot CWs were below the national limit for 396 environmental discharge (500 μg/L; IEAP, 2008). Zn removal can be considered a first-order 397 process. For a SFCW and HSFCW mesocosm at Sacramento, California, the estimated k-rate was 398 28 m/yr (Sinicrope et al., 1992), while for a SFCW treating oil refinery effluent, the k-rate for Zn 399 removal was estimated at 55-121 m/yr at two different loading rates (Gillespie et al., 2000; Kadlec 400 and Wallace, 2009). In a pilot-scale CW treating oil produced water (Kanagy et al., 2008), Zn 401 removal data indicated a k-rate of approximately 15 m/yr. In another pilot-scale SFCW treating 402 ash basin water, Zn removal data suggested k-rates of 21-42 m/yr (Dorman et al., 2009). In this 403 study, the estimated k-rates ranged from 118-146 m/yr (low HRT) to 360-450 m/yr (high HRT) 4044 (Table 3). 0 4 4054 0 5 406 3.3. Phenols 407 Phenol was essentially completely removed in all systems (Fig. 4), as the influent (638 μg/L) was 408 removed to close to non-detect concentrations (0.9-18 μg/L; Tables 1 and 2). First order rate 32 409 coefficients (k) ranged from 152 – 201 m/yr (HRT of 3.16 days) and 96 – 165 m/yr (HRT of 1.37 410 days). It is reported that aerobic biodegradation is the main phenol removal mechanism in CW 411 (Kurzbaum et al., 2010; Rossmann et al., 2012; Stefanakis et al., 2016), though anaerobic 412 degradation is also possible (Kumaran and Paruchuri, 1997). 413 Similarly, high removal rates for phenols are reported in CW studies in the literature, even for 414 higher influent concentrations. Data from a series of pilot-scale HSFCW treating coffee processing 415 wastewater in Brazil (Rossmann et al., 2012) suggests phenol degradation k-rates of 8-17 m/yr. A 416 pilot HSFCW treating cork boiling wastewater in Portugal suggests k-rates between 35-42 m/yr 417 (Gomes et al., 2018). In a series of pilot studies using SFCW (Polprasert et al., 1996), data 418 indicated a phenol degradation k-rate of approximately 100 m/yr. In a study of planted and 419 unplanted SFCW channels in China, k-rates for phenol degradation were estimated at 150 m/yr 420 (planted) and 76 m/yr (unplanted) (Tang et al., 1993). In a study at a petroleum facility, a full-scale 421 SFCW (50 ha) had an estimated k-rate of approximately 100 m/yr for phenol removal (Dong and 422 Lin, 1994). Complete phenols removal and a k rate of 107.2 m/yr was found in a HSFCW system 423 treating groundwater contaminated with phenols and petroleum derivatives (Stefanakis et al., 424 2016). A similar phenols removal k rate of 98.9 m/yr is reported for a HSFCW treating pre-treated 4254 olive mill wastewater (Tatoulis et al., 2017). 2 5 4264 2 6 427 3.4. Nitrogen 428 Ammonia nitrogen (NH4-N) removal (nitrification) occurred in all pilot units at both loading rates 429 (influent of 31,000 μg/L, effluent of 18-20 μg/L; Tables 1 and 2), with corresponding k-rates of 430 233-237 m/yr (Table 3). NH4-N removal was strongly influenced by the presence of aeration (Fig. 431 4). It is likely the actual k-rates are higher, since the systems ran out of NH4-N with essentially 33 432 complete removal. In a similar study, an aerated HSFCW in Germany had an estimated k-rate of 433 1,540 m/yr vs. 2.9 m/yr for a non-aerated system (Table 3; Nivala et al., 2019). A high k-rate of 434 960.6 m/yr was also estimated for an aerated VFCW in the UK operating as tertiary treatment stage 435 (Stefanakis et al., 2019), while an aerated second-stage HSFCW in Oman suggests k-rate of 589 436 m/yr (Stefanakis, 2020b). In both these systems under different climates, complete ammonia 437 removal occurred as well, indicating the positive impact of artificial aeration on aerobic 438 nitrification (Stefanakis et al., 2019). 439 Ammonia removal was also influenced by the presence of vegetation (Tables 1 and 2), since the 440 presence of plants promotes the biofilm development and the related aerobic oxidation of NH4-N 441 by microorganisms (Stefanakis, 2016). Many studies report that plants’ presence favor ammonia 442 removal (Kadlec and Wallace, 2009; Stefanakis et al., 2016; Schultze-Nobre et al., 2017) The 443 shorter duration trial (HRT of 1.37 days) had an estimated k-rate of 39 m/yr, whereas at the longer 444 HRT (3.16 days), the estimated k-rate was 233 m/yr (since the system ran out of NH4-N). Without 445 plants or aeration, k-rates were lower, 13.8 m/yr (HRT of 1.37 days) and 24.4 m/yr (HRT of 3.16 446 days), respectively. Respective k-rate of 22.4 and 39.9 m/yr are reported for HSF systems treating 447 contaminated groundwater and cork boiling wastewater (Stefanakis et al., 2016; Gomes et al., 448 2018). 449 Nitrate (NO3-N) removal (denitrification) also occurred but al lower levels. Nitrate concentrations 450 decreased from influent values, despite the conversion of NH4-N to oxidized forms of nitrogen 451 within the pilot systems. Aeration possibly impacted NO3-N removal, as more NH4-N was 452 converted to oxidized forms of nitrogen, and the aerobic conditions were not conducive to 453 complete denitrification. For systems without vegetation, estimated k-rates with and without 454 aeration were 0.9-2.0 m/yr (HRT of 1.37 days) and 6.0-9.3 m/yr (HRT of 3.16 days), respectively. 34 455 With vegetation, NO3-N removal rates increased (Fig. 4), potentially due to plant uptake 456 considering that ammonia is completely oxidized and nitrate remain the only available nitrogen 457 form for plant uptake, as well as due to the potential presence of more bioavailable carbon in the 458 system, but again, NO3-N removal rates were possibly by aeration. However, the presence of non- 459 aerated sections in the design possibly limited the effect of the aeration in the denitrification extent 460 in the system. The fact that aeration also enhances the microbial growth (Chazarenc et al., 2009; 461 Nivala et al., 2020) indicates that nitrate removal should be a combination of microbial oxidation 462 and plant uptake. Estimated k-rates with and without aeration were 51-73 m/yr (HRT of 1.37 days) 463 and 16-19 m/yr (HRT of 3.16 days). In a similar study, an aerated HSFCW in Germany (Nivala et 464 al., 2019) had an estimated k-rate of 75 m/yr with aeration and 3.9 m/yr without aeration, since 465 NO3-N was not produced in the non-aerated wetland. Similar k-rates of 65.1 and 96.4 m/yr were 466 also estimated for an aerated VFCW in the UK (Stefanakis et al., 2019) and an aerated HSFCW in 467 Oman (Stefanakis, 2020b), with the latter receiving small amounts of raw wastewater (step- 4684 feeding) to increase the carbon availability, hence the higher rate observed. 6 8 4694 6 9 470 3.5. Phosphorus 471 Phosphorus (PO4-P) removal only occurred when vegetation was present in the system, since it is 472 known that PO4-P removal in CW systems is generally lower compared to other pollutants and 473 mostly depends on the adsorption capacity of the substrate media and the uptake by plants to cover 474 their growth needs (Kadlec and Wallace, 2009; García et al., 2010; Stefanakis, 2016). For the trials 475 conducted with vegetation, phosphorus uptake was essentially complete (95-98%) with effluent 476 concentrations close to detection limits. The applied phosphorus loads of 151 and 507 g/m²/yr (at 35 477 the low and high loading rates) greatly exceeds the amount of phosphorus required to support the 36 478 plant biomass cycle (Kadlec and Wallace, 2009). As a result, it is anticipated that the system would 479 saturate with phosphorus in a longer-term study as the plant community matures and the media 4804 gets gradually saturated. 8 0 4814 8 1 482 3.6. Optimum design configuration 483 The tested design proved to be efficient in treating the refinery effluent reaching higher removal 484 rates for most of the tested parameters. The different pilot units revealed the significant role of 485 plants in the overall performance, as the planted bed (RW-SP) showed better removal rates for all 486 parameters than the unplanted unit (RW-S) (Tables 2 and 3), with statistically significant removals 487 (p<0.05) occurring for nitrogen and phosphorus and many heavy metals (Al, Ce, Cr, K, Mo, Sc, 488 Se, Si, Sn, Sr, Ta, W). The performance appears to be further enhanced with the addition of the 489 aerated section (RW-SPA; Tables 2 and 3). Artificial aeration seems to improve (p<0.05) the 490 removal of ammonia nitrogen, Al, As, K, Mn, Sc, and Se compared to the planted non-aerated unit 491 (RW-SP). On the other hand, when the aeration was applied in the unplanted unit (RW-SA), it did 492 not improve the overall performance. Among the two HRT tested, high removals were observed 493 even for the smaller HRT of 1.4 days with slightly higher removal rates compared to the removals 494 under the HRT of 3.2 days and with statistical significance (p<0.05) occurring for nitrate, Al, As, 495 K, Li, Mg, Mn, Ni, Se, Sr, and Ta. 496 The advantage of the tested design arises from the comparison with other wetland designs applied 497 for similar effluent treatment. The surface flow CW is widely applied for such effluents, but it has 498 a significantly higher area demand due to the prolonged HRT required, e.g., 20 days under similar 499 hydraulic load of 0.06 m/d (Mustafa et al., 2018). A VFCW operating at 0.07 m/d and a HRT of 2 37 500 days showed a lower performance for nutrients though receiving a much lower influent 38 501 concentration since it received a secondary refinery effluent (Mustapha et al., 2015; 2018). A 502 HSFCW was also tested for diesel contaminated water at different levels, resulting in optimum 503 removal conditions of 63 days retention time (Al-Baldawi et al., 2014). In a review paper on 504 refinery effluent treatment in CW, the HRT range reported varies between 2.1-90 days, with the 505 majority of the systems being in the range above 10 days (Jia et al., 2020). In the same review 506 study, the positive role of plants and aeration is also indicated, while the HSFCW is reported to be 5075 the design with the better performance. 0 7 5085 0 8 509 4. Conclusions 510 Petroleum refinery effluents represent a major industrial pollution source of significant concern 511 due to its toxic and harmful nature. Established biological methods are not very effective in the 512 remediation of these effluents, while advanced oxidation processes, although effective, are 513 expensive to install and operate. In this study, the technology of Constructed Wetlands was applied 514 for refinery effluent containing heavy metals, phenols and nitrogen as a sustainable nature-based 515 treatment solution. An innovative CW design forming a meandric ‘racetrack’ shape and combining 516 different treatment environments (aerobic and anaerobic) was tested for the first time at pilot-scale. 517 Different construction (non-aerated, artificial aeration, planted and unplanted) and operation 518 (different hydraulic loading rates) parameters were studied to determine the optimum design for 519 effective performance. The first-order removal of metals and the simultaneous removal of phenols 520 and nutrients from the oil refinery wastewater were calculated. The planted unit with artificial 521 aeration managed to reach high removal rates for Fe (96-98%), Mn (38-81%), Al (49-73%), Zn 522 (99-100%), Cu (61-80%), Ni (70-85%), Pb (96-99%) and Cr (60-92%) even at the short HRT of 39 523 1.3 days, while complete removal of phenols and full nitrification was also found. As the calculated 40 524 first-order removal coefficients showed, the overall performance of this novel design exceeded the 525 reported efficiency of the common CW designs and implies the potential to further apply and test 526 it for other industrial effluents. 527 528 Authorship contribution statement 529 Mozaffari M.S: Conceptualization, Writing - review, Visualization, Investigation, Data curation. 530 Shafiepour E.: Project administration. Mirbagheri S.A.: Resources, Funding acquisition 531 Rakhshandehroo G.: Methodology, Funding acquisition. Wallace S.: Conceptualization, Formal 532 analysis, Data curation. Stefanakis A.I.: Formal analysis, Data curation and analysis, Writing - 5335 original draft, Writing - review 3 3 5345 3 4 535 Acknowledgements 536 This study work was supported by the Civil Engineering Department, Shiraz University and the 5375 Khajeh Nasir Toosi University of Technology. 3 7 5385 3 8 539 References 540 APHA, 1999. Standard Methods for the Examination of Water and Wastewater Part. American 541 Water Works Association (AWWA). 542 Al-Baldawi, I.A.W., Abdullah, S.R.S., Hasan, H.A.S., Suja, F., Anuar, N., Mushrifah, I., 2014. 543 Optimized conditions for phytoremediation of diesel by Scirpus grossus in horizontal 41 544 subsurface flow constructed wetlands (HSFCWs) using response surface methodology. J. 545 Environ. Manage. 140, 152-159. 42 546 Aslam, M.M., Mali, M., Baig, M.A., Qazi, I.A., Iqbal, J., 2007. Treatment performances of 547 compost-based and gravel-based vertical flow wetlands operated identically for refinery 548 wastewater treatment in Pakistan. Ecol. Eng. 30, 34-42, 549 https://doi.org/10.1016/j.ecoleng.2007.01.002 550 Bidhendi, N.G.R., Mehrdadi, N., Mohammadnejad, S., 2010. Water and wastewater minimization 551 in Tehran oil refinery using water pinch analysis. Int. J. Environ. Res. 4(4), 583-594, 552 doi:10.22059/IJER.2010.244 553 Boog, J., 2013. Effect of the aeration scheme on the treatment performance of intensified treatment 554 wetland systems. TU Bergakademie Freiberg: Freiberg, Germany. 555 Boog, J., Nivala, J., Aubron, T., Wallace, S., van Afferden, M., Müller, R.A., 2014. Hydraulic 556 characterization and optimization of total nitrogen removal in an aerated vertical subsurface 557 flow treatment wetland. Bioresour. Technol. 162, 166-74, 558 https://doi.org/10.1016/j.biortech.2014.03.100 559 Chazarenc, F.; Gagnon, V.; Comeau, Y.; Brisson, J. Effect of plant and artificial aeration on solids 560 accumulation and biological activities in constructed wetlands. Ecol. Eng. 2009, 35, 1005– 561 1010. 562 Del Rio, L., Hadwin, A.K.M., Pinto, L.J., MacKinnon, M.D., Moore, M.M., 2006. Degradation of 563 naphthenic acids by sediment micro‐ organisms. J. Appl. Microbiol. 101(5), 1049-1061, 564 https://doi.org/10.1111/j.1365-2672.2006.03005.x 565 DeVolder, P.S., Brown, S.L., Hesterberg, D., Pandya, K., 2003. Metal bioavailability and 566 speciation in a wetland tailings repository amended with biosolids compost, wood ash, and 567 sulfate. J. Environ. Qual. 32(3), 851-864, https://doi.org/10.2134/jeq2003.8510 43 568 Dong, K., Lin, C., 1994. Treatment of petrochemical wastewater by the system of wetlands and 569 oxidation ponds and engineering design of the system. Guangzhou, P.R. China: IWA. 570 Dorman, L., Castle, J.W., Rodgers, Jr., J.H., 2009. Performance of a pilot-scale constructed 571 wetland system for treating simulated ash basin water. Chemosphere 75(7), 939-47, 572 https://doi.org/10.1016/j.chemosphere.2009.01.012 573 García, J., Rousseau, D.P.L., Morató, J., Lesage, E., Matamoros, V., Bayona, J.M., 2010. 574 Contaminant removal processes in subsurface-flow constructed wetlands: a review. Crit. Rev. 575 Environ. Sci. Technol. 40, 561-661, https://doi.org/10.1016/j.chemosphere.2009.01.012 576 Gensemer, R.W., Playle, R.C., 1999. The bioavailability and toxicity of aluminum in aquatic 577 environments. Crit. Rev. Environ. Sci. Technol. 29(4), 315-450, 578 https://doi.org/10.1080/10643389991259245 579 Gholipour, A., Zahabi, H., Stefanakis, A.I., 2020. A novel pilot and full-scale constructed wetland 580 study for glass industry wastewater treatment. Chemosphere 247, 125966, 581 https://doi.org/10.1016/j.chemosphere.2020.125966 582 Gillespie Jr, W.B., Hawkins, W.B., Rodgers Jr, J.H., Cano, M.L., Dorn, P.B., 2000. Transfers and 583 transformations of zinc in constructed wetlands: Mitigation of a refinery effluent. Ecol. Eng. 584 14(3), 279-292, https://doi.org/10.1006/eesa.1999.1762 585 Gomes, A.C., Silva, L., Albuquerque, A., Simões, R., Stefanakis, A.I., 2018. Investigation of lab- 586 scale horizontal subsurface flow constructed wetlands treating industrial cork boiling 587 wastewater. Chemosphere 207, 430-439, https://doi.org/10.1016/j.chemosphere.2018.05.123 588 Gorito, A.M., Ribeiro, A.R., Almeida, C.M.R., Silva, A.M.T., 2017. A review on the application 589 of constructed wetlands for the removal of priority substances and contaminants of emerging 44 590 concern listed in recently launched EU legislation. Environ. Pollut. 227, 428-443, 591 https://doi.org/10.1016/j.envpol.2017.04.060 592 Hallberg, K.B., Johnson, D.B., 2005. Biological manganese removal from acid mine drainage in 593 constructed wetlands and prototype bioreactors. Sci. Total Environ. 338, 115-124, 594 https://doi.org/10.1016/j.scitotenv.2004.09.011 595 Hedin, R.S, Nairn, R.W., 1993. Contaminant removal capabilities of wetlands constructed to treat 596 coal mine drainage. In: Constructed Wetlands for Water Quality Improvement, G.A. Moshiri, 597 Editor. CRC Press: Boca Raton, Florida, p. 187-196. 598 Hedin, R.S., Nairn, R.W., Kleinman, R.L.P., 1994. Passive treatment of coal mine drainage with 599 limestone. U.S. Bureau of Mines: Pittsburgh, Pennsylvania, p. 35. 600 IEPA, 2008. Environmental regulations and standards of Iran. Iran Global Environmental Law. 601 Tehran, pp. 234-39. 602 Jain, M., Majumder, A., Ghosal, P.S., Gupta, A.K., 2020. A review on treatment of petroleum 603 refinery and petrochemical plant wastewater: A special emphasis on constructed wetlands. J. 604 Environ. Manage. 272, 111057, https://doi.org/10.1016/j.jenvman.2020.111057 605 Ji, G.D., Sun, T.H., Ni, J.R., 2007. Surface flow constructed wetland for heavy oil-produced water 606 treatment. Bioresour. Technol. 98, 436–441, https://doi.org/10.1016/j.biortech.2006.01.017 607 Johnson, B.M., Kanagy, L.E., Rodgers Jr, J.H., Castle, J.W., 2008. Feasibility of a pilot-scale 608 hybrid constructed wetland treatment system for simulated natural gas storage produced waters. 609 Environ. Geosci. 15(3), 91-104, https://doi.org/10.1306/eg.06220707004 610 Kadlec, R.H., Wallace S.D., 2009. Treatment Wetlands, Second Edition. Boca Raton, Florida: 611 CRC Press. 45 612 Kadlec, R.H., Srinivasan, K., 1995. Wetland Treatment of Oil and Gas Well Wastewaters. U.S. 613 Department of Energy: Bartlesville, Oklahoma. 614 Kanagy, L., Johnson, B.M., Castle, J.W., Rodgers Jr, J.H., 2008. Design and performance of a 615 pilot-scale constructed wetland treatment system for natural gas storage produced water. 616 Bioresour. Technol. 99(6). 1877-1885, https://doi.org/10.1016/j.biortech.2007.03.059 617 Kumaran, P., Paruchuri, Y.L., 1997. Kinetics of phenol biotransformation. Water Res. 31(1), 11- 618 22, https://doi.org/10.1016/S0043-1354(99)80001-3 619 Kurzbaum, E., Kirzhner, F., Sela, S., Zimmels, Y., Armon, R., 2010. Efficiency of phenol 620 biodegradation by planktonic Pseudomonas pseudoalcaligenes (a constructed wetland isolate) 621 vs. root and gravel biofilm. Water Res, 44(17), 5021-5031. 622 Langwaldt, J., Puhakka, J., 2000. On-site biological remediation of contaminated groundwater: a 623 review. Environ. Pollut. 2000. 107(2), 187-197, https://doi.org/10.1016/S0269-7491(99)00137- 624 2 625 Manyin, T., Williams, F.M., Stark, L.R., 1997. Effects of iron concentration and flow rate on 626 treatment of coal mine drainage in wetland mesocosms: An experimental approach to sizing of 627 constructed wetlands. Ecol. Eng. 9(3-4), 171-186, https://doi.org/10.1016/S0925- 628 8574(97)10005-2 629 Mustafa, A., Azim, M.K., Raza, Z., Kori, J.A., 2018. BTEX removal in a modified free water 630 surface wetland. Chem. Eng. J. 333, 451–455. 631 Mustapha, H.I., van Bruggen, J.J.A., Lens, P.N.L., 2015. Vertical subsurface flow constructed 632 wetlands for polishing secondary Kaduna refinery wastewater in Nigeria. Ecological 633 Engineering 84, 588–595. 46 634 Mustapha, H.I., van Bruggen, J.J.A., Lens, P.N.L., 2018. Fate of heavy metals in vertical 635 subsurface flow constructed wetlands treating secondary treated petroleum refinery wastewater 636 in Kaduna, Nigeria. Int. J. Phytoremed. 20(1), 44-53, 637 https://doi.org/10.1080/15226514.2017.1337062 638 Nivala, J., Kahl, S., Boog, J., van Afferden, M., Reemtsma, T., Müller, R., 2019. Dynamics of 639 emerging organic contaminant removal in conventional and intensified subsurface flow 640 treatment wetlands. Sci. Total Environ. 649, 1144-1156, 641 https://doi.org/10.1016/j.scitotenv.2018.08.339 642 Nivala J., Murphy, C., Freeman, A., 2020. Recent Advances in the Application, Design, and 643 Operations & Maintenance of Aerated Treatment Wetlands. Water 12(4), 1188; 644 https://doi.org/10.3390/w12041188 645 Nyquist, J., Greger, M., 2009. A field study of constructed wetlands for preventing and treating 646 acid mine drainage. Ecol. Eng. 35(5), 630-642, https://doi.org/10.1016/j.ecoleng.2008.10.018 647 PIRAMID Consortium, 2003. Engineering guidelines for the passive remediation of acidic and/or 648 metalliferous mine drainage and similar wastewaters. European Commission 5th Framework 649 RTD Project no. EVK1-CT-1999-000021, Passive in-situ remediation of acidic mine-industrial 650 drainage, University of Newcastle Upon Tyne: Newcastle Upon Tyne, UK. p.166. 651 Polprasert, C., Dan, N.P., Thayalakumaran, N., 1996. Application of constructed wetlands to treat 652 some toxic wastewaters under tropical conditions. Water Sci. Technol. 34(11), 165-171, 653 https://doi.org/10.1016/S0273-1223(96)00834-7. 654 Qureshi, N., Annous, B.A., Ezeji, T.C., Karcher, P., Maddox, I.S., 2005. Biofilm reactors for 655 industrial bioconversion processes: employing potential of enhanced reaction rates. Microbial 656 Cell Factories 4, 24, https://doi.org/10.1186/1475-2859-4-24. 47 657 Ramírez, S., Torrealba, G., Lameda-Cuicas, E., Molina-Quintero, L., Stefanakis, A.I., 2019. 658 Investigation of pilot-scale Constructed Wetlands treating simulated pre-treated tannery 659 wastewater under tropical climate. Chemosphere 234, 496-504, 660 https://doi.org/10.1016/j.chemosphere.2019.06.081 661 Ramond, J-B., Welz, P.J., Cowan, D.A., Burton, S.G., 2012. Microbial community structure 662 stability, a key parameter in monitoring the development of constructed wetland mesocosms 663 during start-up. Research in Microbiology 163(1), 28-35. 664 Rossmann, M., de Matos, A.T., Abreu, E.C., Silva, F.F., Borges, A.C., 2012. Performance of 665 constructed wetlands in the treatment of aerated coffee processing wastewater: Removal of 666 nutrients and phenolic compounds. Ecol. Eng. 2012. 49, 264-269, 667 https://doi.org/10.1016/j.ecoleng.2012.08.017 668 Schultze-Nobre, L., Wiessner, A., Bartsch, C., Paschke, H., Stefanakis, A.I., Aylward, L.A., 669 Kuschk, P., 2017. Removal of dimethylphenols and ammonium in laboratory-scale horizontal 670 subsurface flow Constructed Wetlands. Eng. Life Sci. 17(12), 1224-1233, 671 https://doi.org/10.1002/elsc.201600213 672 Sheoran, A.S., Sheoran, V., 2006. Heavy metal removal mechanism of acid mine drainage in 673 wetlands: A critical review. Mineral Eng. 19, 105-116, 674 https://doi.org/10.1016/j.mineng.2005.08.006 675 Sikora, F.J., Behrends, L.L., Brodie, G.A., Taylor, H.N., 2000. Design criteria and required 676 chemistry for removing manganese in acid mine drainage using subsurface flow wetlands. 677 Water Environ. Res. 72(5), 536-544, https://doi.org/10.2175/106143000X138111 48 678 Sinicrope, T.L., Langis, R., Gersberg, R.M., Busnardo, M.J., Zedler, J.B., 1992. Metal removal by 679 wetlands mesocosms subjected to different hydroperiods. Ecol. Eng. 1(4), 309-322, 680 https://doi.org/10.1016/0925-8574(92)90013-R 681 Stefanakis, A.I., 2016. Constructed Wetlands: description and benefits of an eco-tech water 682 treatment system. In: McKeown, A.E., Bugyi, G. (Eds.), Impact of Water Pollution on Human 683 Health and Environmental Sustainability. IGI Global, Hershey PA, USA, pp. 281-303, 684 doi:10.4018/978-1-4666-9559-7.ch012 685 Stefanakis, A.I., 2018. Constructed wetlands for industrial wastewater treatment. John Wiley & 686 Sons Ltd, Chichester, UK, doi:10.1002/9781119268376 687 Stefanakis, A.I., 2019. The role of constructed wetlands as green infrastructure for sustainable 688 urban water management. Sustainability 11(24), 6981, https://doi.org/10.3390/su11246981 689 Stefanakis, A.I., 2020a. The fate of MTBE and BTEX in Constructed Wetlands. Appl. Sci. 10, 690 127; doi:10.3390/app10010127. 691 Stefanakis, A.I., 2020b. Constructed wetlands for sustainable wastewater treatment in hot and arid 692 climates: opportunities, challenges and case studies in the Middle East. Water 12(6), 1665, 693 https://doi.org/10.3390/w12061665 694 Stefanakis, A.I., Seeger, E., Dorer, C., Sinke, A., Thullner, M., 2016. Performance of pilot-scale 695 horizontal subsurface flow constructed wetlands treating groundwater contaminated with 696 phenols and petroleum derivatives. Ecol. Eng. 95, 514-526, 697 https://doi.org/10.1016/j.ecoleng.2016.06.105 698 Stefanakis, A.I., Prigent, S., Breuer, R., 2018. Integrated produced water management in a desert 699 oilfield using wetland technology and innovative reuse practices. in: Stefanakis, A.I. (ed.), 49 700 Constructed wetlands for industrial wastewater treatment. John Wiley & Sons Ltd, Chichester, 701 UK, pp. 25-42, https://doi.org/10.1002/9781119268376.ch1 702 Stefanakis, A.I., Bardiau, M., Silva, D., Taylor, H., 2019. Presence of bacteria and bacteriophages 703 in full-scale trickling filters and an aerated constructed wetland. Sci. Total Environ. 659, 1135– 704 1145, https://doi.org/10.1016/j.scitotenv.2018.12.415 705 Tang, S.Y., Lu, X.W., 1993. The use of Eichhornia crassipes to cleanse oil refinery wastewater in 706 China. Ecol. Eng. 2(3), p. 243-251, https://doi.org/10.1016/0925-8574(93)90017-A 707 Tarutis, W.J., Stark, L.R., Williams, F.M., 1999. Sizing and performance estimation of coal mine 708 drainage wetlands. Ecol. Eng. 12(4), 353-372, https://doi.org/10.1016/S0925-8574(98)00114- 709 1 710 Tatoulis, T., Stefanakis, A.I., Frontistis, Z., Akratos, C.S., Tekerlekopoulou, A.G., Mantzavinos, 711 D., Vayenas, D.V., 2016. Treatment of table olive washing water using trickling filters, 712 constructed wetlands and electrooxidation. Environ. Sci. Pollut. Res. 1-8, doi:10.1007/s11356- 713 016-7058-6. 714 Vymazal, J., Krasa, P., 2003. Distribution of Mn, Al, Cu and Zn in a constructed wetland receiving 715 municipal sewage. Water Sci. Technol. 48(5), 299-305, https://doi.org/10.2166/wst.2003.0336 716 Wallace, S.D., 2002. Method for removing pollutants from water. United States. 717 Wallace, S., Schmidt, M., Larson, E., 2011. Long term hydrocarbon removal using treatment 718 wetlands. SPE Annual Technical Conference and Exhibition, Denver, Colorado, USA, SPE- 719 145797-MS, https://doi.org/10.2118/145797-MS 720 Wu, S., Wallace, S., Brix, H., Kuschk, P., Kirui, W.K., Masi, F., Dong, R., 2015. Treatment of 721 industrial effluents in constructed wetlands: Challenges, operational strategies and overall 722 performance. Environ. Pollut. 201, 07-120, https://doi.org/10.1016/j.envpol.2015.03.006 50 723 Younger, P.L., Banwart, S.A., Hedin, R.S., 2002. Mine Water: Hydrology, Pollution, 724 Remediation. Kluwer Academic Publishers: London, United Kingdom. doi:10.1007/978-94- 725 010-0610-1 726 Xin, J., Tang, J., Liu, Y., Zhang, Y., Tian, R., 2019. Pre-aeration of the rhizosphere offers potential 727 for phytoremediation of heavy metal-contaminated wetlands. J. Hazard. Mater. 374, 437-446, 728 https://doi.org/10.1016/j.jhazmat.2019.04.010 51 Manuscript Click here to view linked References 1 Hydraulic characterization and removal of metals and nutrients in an aerated 2 horizontal subsurface flow “racetrack” wetland treating primary-treated oil 3 industry effluent 4 Mohammad-Hosein Mozaffari¹, Ehsan Shafiepour², Seyed Ahmad Mirbagheri², Gholamreza 5 Rakhshandehroo³, Scott Wallace4, Alexandros I. Stefanakis5* 6 ¹Civil Engineering Department, Montana State University, Montana, USA 7 ²Khajeh Nasir Toosi University of Technology, Tehran, Iran, 8 ³Department of Civil and Environmental Engineering, Shiraz University, Shiraz, Iran. 9 4Naturally Wallace Consulting LLC, Stillwater, MN 55082, USA 10 5Laboratory of Environmental Engineering and Management, School of Environmental 11 Engineering, Technical University of Crete, 73100 Chania, Greece 12 * Corresponding author: astefanakis@enveng.tuc.gr 13 14 Highlights 15 • Novel ‘racetrack’ constructed wetland design tested for the first time 16 • Treatment of oil refinery effluent rich in heavy metals, phenols and nutrients 17 • Positive role of plants and artificial aeration found, even at low retention time of 1.3 days 18 • High heavy metals removal (up to 98%), complete degradation of phenols and ammonia 19 • Higher k-rates calculated than literature findings for other wetland systems 20 21 Abstract 1 22 Constructed wetlands (CW) are an attractive technology due to their operational simplicity and 23 low life-cycle cost. It has been applied for refinery effluent treatment but mostly single-stage 24 designs (e.g., vertical or horizontal flow) have been tested. However, to achieve a good treatment 25 efficiency for industrial effluents, different treatment conditions (both aerobic and anaerobic) are 26 needed. This means that hybrid CW systems are typically required with a respectively increased 27 area demand. In addition, a strong aerobic environment that facilitates the formation of iron, 28 manganese, zinc and aluminum precipitates cannot be established with passive wetland systems, 29 while the role of these oxyhydroxide compounds in the further co-precipitation and removal of 30 heavy metals such as copper, nickel, lead, and chromium that can simplify the overall treatment of 31 industrial wastewaters is poorly understood in CW. Therefore, this study tests for the first time an 32 innovative CW design that combines an artificially aerated section with a non-aerated section in a 33 single unit applied for oil refinery wastewater treatment. Four pilot units were tested with different 34 design (i.e., planted/unplanted, aerated/non-aerated) and operational (two different hydraulic 35 loading rates) characteristics to estimate the role of plants and artificial aeration and to identify the 36 optimum design configuration. The pilot units received a primary refinery effluent, i.e., after 37 passing through a dissolved air flotation unit. The first-order removal of heavy metals under 38 aerobic conditions is evaluated, along with the removal of phenols and nutrients. High removal 39 rates for Fe (96-98%), Mn (38-81%), Al (49-73%), and Zn (99-100%) generally as oxyhydroxide 40 precipitates were found, while removal of Cu (61-80%), Ni (70-85%), Pb (96-99%) and Cr (60- 41 92%) under aerobic conditions was also observed, likely through co-precipitation. Complete 42 phenols and ammonia nitrogen removal was also found. The first-order rate coefficient (k) 43 calculated from the collected data demonstrates that the tested CW represents an advanced wetland 44 design reaching higher removal rates at a smaller area demand than the common CW systems. 2 45 46 Keywords: aerated wetland; constructed wetland; refinery effluent; phenols; heavy metals; tracer 47 test 48 49 1. Introduction 50 Polluted water from industrial sites represents a major environmental threat for ecosystems and 51 public health by virtue of historic operational practices, incidents and leakages during storage and 52 transport (Langwaldt and Puhakka, 2000; Stefanakis et al., 2020a). These wastewaters typically 53 have a complex composition with a large variety of pollutants of different nature and properties 54 such as heavy metals, organic matter, nutrients, solids, color, salinity, amid of other inorganic and 55 toxic compounds (Wu et al., 2015; Stefanakis, 2018; Ramirez et al., 2019). Among them, heavy 56 metals are considered one the most problematic. Many industrial wastewaters have varying levels 57 of metals that derive from their operation and require further treatment prior the release to the 58 environment or reuse at the operation sites. In particular, heavy metals are frequently found in 59 effluents from refineries and generally the petrochemical industry (Wu et al., 2015; Mustapha et 60 al., 2018; Stefanakis et al., 2018; Jain et al., 2020). 61 Various technologies and processes have been applied for the treatment of refinery effluents, e.g., 62 membrane bioreactors, activated sludge, anaerobic reactors, ozonation etc. (Stefanakis, 2020a; Jain 63 et al., 2020). The green technology of Constructed wetlands (CW) has also been used for petroleum 64 derived water treatment though the respective applications are limited (Ji et al., 2007; Johnson et 65 al., 2008; Wallace et al., 2011; Stefanakis et al., 2016; Stefanakis, 2020a). The positive aspects of 66 CW include lower investment for infrastructure, significantly lower operation and maintenance 67 cost, ecological character, tolerance to flowrate variations, no use of chemicals for the treatment, 3 68 among others (Kadlec and Wallace, 2009; Stefanakis, 2019). In addition, CW can handle varying 69 contaminant loads and treat many constituents of concern at the same time and more effectively 70 than some physical and chemical processes (Stefanakis, 2016; Gorito et al., 2017). These 71 advantages account for the growing interest and increasing number of CW applications for the 72 treatment of various industrial wastewater sources such as from the chemical and petrochemical 73 industry (Wallace et al., 2011; Stefanakis et al., 2016; Stefanakis, 2020a), the oil and gas 74 exploration (Stefanakis et al., 2018; Jain et al., 2020; Stefanakis, 2020b), the food industry such as 75 dairy farms and olive mills (Tatoulis et al., 2017; Wu et al., 2015), and agro-industries such as 76 glass industry, tanneries, cork processing wastewater (Gholipour et al., 2020; Gomes et al., 2018; 77 Ramirez et al., 2019), among others. 78 The application of CW technology for refinery effluent treatment is not widely reported, although 79 current knowledge indicates that CW can be specifically designed and built for the removal of 80 target-constituents from refinery effluents. A few studies on vertical flow CW (VFCW) and 81 horizontal subsurface flow CW designs (HSFCW) are also available in the international literature 82 (Gillespie et al., 2000; Stefanakis et al., 2016; Mustapha et al., 2018; Jain et al., 2020). HSFCW 83 are inherently oxygen-transfer limited systems, thus, anaerobic digestion (methanogenesis) 84 becomes a salient contributor to the removal of the various constituents (Kadlec and Wallace, 85 2009). On the other hand, artificial aeration in CW is an advanced design that increases the 86 phytoremediation potential of CW for heavy metals removal such as Cu+2 (Xin et al., 2019), while 87 it also enhances the biodegradation rate under an aerobic environment (Del Rio et al., 2006; 88 Stefanakis et al., 2019). 89 However, the vast majority of the published literature studies single CW units that utilize only one 90 wetland design (e.g., vertical or horizontal or surface flow), while the aerated CW has barely been 4 91 tested for refinery effluent treatment. This means that a CW system for refinery effluent typically 92 has a high areal footprint. However, the refinery effluent composition would require a combination 93 of different environments, i.e., aerobic and anaerobic conditions, to achieve effective removal rates 94 for the various pollutants especially for nitrogen and heavy metals. This could potentially be done 95 with the use of hybrid (i.e., multistage) CW systems but this would further increase the land area 96 demand. Therefore, the present study proposes and investigates an advanced CW design as a new 97 innovative solution that combines in the same unit different treatment conditions necessary for the 98 effective removal of pollutants of diverse nature aiming at providing a wetland treatment solution 99 with smaller area demand than the existing CW designs. Therefore, the main research goal of this 100 study is to test for the first time this innovative CW design at pilot scale for the effective treatment 101 of process waters from a refinery. The specific goals is to determine the role of plants, substrate, 102 hydraulic retention time and aeration in treating the refinery effluent with focus on the load 1031 removal of a series of heavy metals and nutrients. 0 3 1041 0 4 105 2. Materials and methods 106 2.1. Study location 107 Four pilot-scale HSFCW units (planted and unplanted) with an innovative design were built 108 outdoors at the Khajeh-Nasir Toosi University (35°45'48.0"N 51°24'33.9"E) with different design 109 characteristics, i.e., planted, unplanted (i.e., sand filters), with and without artificial aeration. 110 Wastewater volume was collected from the Tehran oil refinery effluent. This facility, located 111 southern of Tehran City (35°32'24.7"N 51°25'37.7"E), is the second largest refinery in Iran and is 112 one of the oldest refineries with an oil refining capacity of approximately 110,000 barrels per day, 113 and a water consumption of nearly 505 m³/h (Bidhendi et al., 2010). It is considered as the best 5 114 performing refinery for the sake of Environmental Protection Act in Iran. The current wastewater 115 treatment plant of the refinery consists of the DAF unit, followed by an aeration tank, a 116 sedimentation tank and sand filtration unit, while the final effluent is discharged to agricultural 117 fields or recycled in the industrial process. The effluent of the dissolved air flotation (DAF) unit 118 was used for this study, since the DAF system achieves high recovery rates of the residual oil and 1191 it is preferred to remain in operation. 1 9 1201 2 0 121 2.1. Pilot-scale CW design 122 The base unit of the innovative design (called “Racetrack wetland”) of each CW was fabricated 123 out of clear acrylic plastic with a total flow path length of 12.8 m, mean channel width of 7.65 cm, 124 and an effective depth of 0.2 m. The flow path was wrapped in a spiral configuration (Fig. 1a), 125 such that each pilot cell required only 0.86 m² of area. The inflow point was at the centre of the 126 unit and the water followed a meandric subsurface flow path, with the outflow point at one corner 127 of the unit (Figs. 1 and 2). The structural design of the tested ‘Racetrack Wetland’ (RW) system 128 furnishes two different environments (Fig. 2): a planted section without aeration (total length of 129 planted section 6.3 m) and an unplanted section with aeration (total length of unplanted section 5.7 130 m). The aerated unplanted section constituted 44% of the total wetland area and included irrigation 131 tubes in the center of the flow path with specific drippers that provided 0.6 L/min air flowrate at 132 air pressure of 100 mbar. Aeration in the CW was based on previous systems used in North 133 America and Europe (Wallace, 2002), using a 0.175 kW air blower capable of delivering up to 134 626.8 L/m²/min. The unplanted sections were filled with a graded, washed silica sand (size 0.8 - 135 1.2 mm, porosity 24%). The same media was used in the planted sections, but with the addition of 136 5% clay on the top 2 cm, to allow for the plant establishment in the silica soil. The inlet section at 6 137 the center of the unit had dimensions of 26x26 cm, while the outlet section had a length of 80 cm 138 and width 15.5 cm, while coarser gravel (6-7 mm) was used at the inlet and outlet sections to 139 facilitate the flow distribution (Fig. 3). Cyperus alternifolius plants of approx. 1 week old were 140 harvested from local water courses at the distance of 35 kilometers from Tehran Oil Refinery and 141 then immediately planted in the planted section of the CW design, covering 56% of the total 1421 wetland area. 4 2 1431 4 3 7 1441 4 4 Figure 1. (a) Racetrack wetland configuration with aeration (base unit without substrate or 1451 4 5 146 vegetation), and (b) the unplanted and planted sections, the coarse gravel at the inlet and outlet 147 sections, and the influent and effluent sampling points. 148148 8 149149 150 Figure 2. Schematic representation (top view) of the planted and non-planted sections of the 1511 racetrack wetland unit and the meandric flow path. 5 1 1521 5 2 153 Four similar pilot RW units were built and operated in parallel to examine the effects of plant, 154 substrate and aeration in the wetland performance, distinctly, with the following configuration: 155 one pilot unit only with substrate (RW-S), one pilot unit with substrate and forced bed aeration 156 (RW-SA), one pilot unit with substrate planted with Cyperus alternifolius (RW-SP), and one pilot 1571 unit with substrate planted with Cyperus alternifolius and with forced bed aeration (RW-SPA). 5 7 1581 5 8 9 159 2.2. Pilot-scale CW operation 10 160 Once the pilot RW units were built and planted, a solution of 10 mL of a natural liquid fertilizer 161 (NPK 7-4-7, Iranian plant company) mixed with 100 mL of tap water was added to each RW twice 162 per week to cover the nutrients requirement for the initial growth period of the plants and biofilm 163 (Ramond et al., 2012). The feeding with the NPK solution lasted the first 3 months after planting, 164 in order to assist plant adoption to the new environment and sufficient penetration depth of plant 165 roots into the substrate (Qureshi et al., 2005). After that start-up period, tracer tests were carried 166 out for a period of 2 weeks. Once the tracer tests were completed, the refinery effluent was 167 collected in a large tank that fed the CW units and the feeding of the pilot CWs started. 168 The first 3 weeks of operation with the refinery effluent were considered a necessary period for 169 the system to adopt to the feed water after the start-up period, hence no samples were taken (Kadlec 170 and Wallace, 2009; Ramirez et al., 2019). After that period, another batch of refinery effluent was 171 collected and fed, and the sampling and monitoring started. Two flowrates, i.e., 0.20 and 0.67 172 mL/s, were selected corresponding to the target hydraulic retention times (HRT) of 4 days and 1 173 day, respectively. Tracer tests were used to determine more accurately the actual HRT at the two 174 different flowrates. The low flowrate (0.2 mL/s) was first applied to the four CWs to evaluate their 175 removal performance. Afterwards, the high low flowrate (0.67 ml/s) was applied and the samples 176 taking started again after a week to allow for the adoption of the system to the new flowrate 1771 (Schultze-Nobre et al., 2017; Tatoulis et al., 2017; Gomes et al., 2018). 7 7 1781 7 8 179 2.3. Sampling and analytical methods 180 Samples were taken after each HRT cycle, i.e., every 3.2 days for operation period under the first 181 HRT (HLR of 0.2 mL/s) and every 1.3 days for operation period under the second HRT (HLR of 182 0.67 mL/s) and a total of 18 samplings were carried out per pilot unit (9 per HRT). The total 11 183 duration of the experiment was 6 months. Wastewater samples were collected from the influent 184 and effluent points (Fig. 1b) and analyzed immediately to determine the following parameters: pH, 185 Al (aluminum), As (arsenic), Ce (Cerium), Cr (Chromium), Cu (Copper), Fe (iron), K (potassium), 186 Li (lithium), Mg (Magnesium), Mn (Manganese), Mo (Molybdenum), Ni (Nickel), Pb (lead), Sc 187 (Scandium), Se (Selenium), Si (Silicon), Sn (Tin), Sr (Strontium), Ta (Tantalum), Th (Thorium), 188 V (Vanadium), W (Tungsten), Zn (Zinc), total phenols, ammonia nitrogen (NH4-N), nitrate (NO3- 189 N) and orthophosphate (PO4-P). Heavy metals were measured via inductively coupled plasma– 190 mass spectrometry (ICP–MS) according to Standard Method 1640 EPA (APHA, 1999). Nitrogen 191 species (NH4-N, NO3-N) were measured via spectrophotometry (T80 UV, PG instrument LTD) 192 according to Standard Method 4500-NO -3 N and ASTM D1426 for NH4-N. Orthophosphate (PO4- 193 P) were measured via spectrophotometry (T80 UV, PG instrument LTD) according to Standard 194 Method 4500-P-C. Total phenols was measured via spectrophotometry (T80 UV, PG instrument 195 LTD) using Standard Method ASTM D1783, and pH was determined by a VIVOSUN PH Meter 196 Digital PH Tester Pen for Water. Conductivity was measured with a conductivity meter (MS1- 1971 IMP403-24) and converted to salinity. Tracer tests were conducted using NaCl. 9 7 1981 9 8 199 2.4. Calculations and statistical analysis 200 To estimate the removal performance, the removal rates of the various parameters were calculated, 201 assuming that the rate coefficients can be described by the means of a first-order removal laws. 202 Considering the flowrates, the HRT, the CW surface area and the concentrations of the various 203 parameters, the first-order reaction rate coefficients (k-rates) for wastewater applied and treated in 204 HSFCW were calculated based on the equation (Tarutis et al., 1999; Kadlec and Wallace, 2009): 205 k = (Q x (lnCin – lnCout))/A 12 206 where, Q is the inflow (m³/day), A the surface area (m²), Cin the influent concentration (mg/L), 207 Cout the effluent concentration (mg/L) and k the first-order rate coefficient (m/d). 208 First-order reaction rate coefficients (k-rates) were calculated from the observed removals and 209 tracer studies (number of the tanks in series) and were compared to those previously reported in 210 the literature. The nominal HRT was calculated based on the void volume of the saturated media 211 without considering the inlet and outlet sections of the units. 212 Statistical analysis was performed using the SPSS software (SPSSInc., Chicago, IL, USA; Version 213 12.0). The data was analysed through one-way analysis of variance (ANOVA) at the 95% 214 significance level (p ≤ 0.05) to compare the performance of the CW units in the removal of the 215 various parameters. Post Hoc pair comparisons were also performed to test equal variations, using 216 Tukey’s honestly significant difference test. Homogeneity of variance tests (Levene) were 2172 bypassed, since the number of data points for each group was the same. 1 7 2182 1 8 219 3. Results and discussion 220 3.1. Tracer experiment results 221 Prior to start measuring the treatment performance, tracer experiments were performed in the RW- 222 SP unit (non-aerated, planted system) at the two different flowrates, i.e., 0.67 and 0.2 mL/s that 223 correspond to the target HRT of 1 day and 4 days, respectively. Sodium chloride (NaCl) was used 224 as a tracer at 1,000 mg/L concentration. Electrical conductivity was measured to obtain the 225 response curve. Tracer data was analyzed using the gamma function distribution method (Kadlec 226 and Wallace, 2009). The actual HRT values were 1.37 and 3.16 days, respectively, with the 227 associated number of tanks-in-series (NTIS) of 6.4 and 12.32 (Fig. 3). Tracer recoveries were 70% 228 and 51% for the high and low flowrate, respectively. 13 2292 2 9 Figure 3. Tracer response curve for the targeted 1-day and 4-days HRT. 2302 3 0 2312 3 1 232 Testing the system at the two different flowrates indicated different hydraulic performances. While 233 the variation in the HRT was roughly proportional to the applied flowrate and more or less 234 expected, the degree of internal mixing (NTIS) shifted from 12.32 (HRT of 3.16 days) to 6.4 (HRT 235 of 1.37 days). This was likely caused by higher flow velocities at the higher flowrate, resulting in 236 a greater degree of internal mixing, despite the length to width ratio of the system being 14 237 approximately 7.7. As a point of reference, the reported mean NTIS value for HSFCW over 37 238 tracer studies is 11.0 (Kadlec and Wallace, 2009). It is expected that the operation of the aeration 15 239 system would increase the internal mixing (Boog, 2013; Boog et al., 2014). A comparable aerated 240 HSFCW at an experimental facility in Germany had an NTIS value of 4.5 while continually aerated 2412 (Boog, 2013), although the system had a length to width ratio of 3.9. 4 1 2422 4 2 243 3.1. Overall treatment performance 244 Table 1 shows the influent and effluent values and the respective removal rates of the various 245 parameters at the flowrate of 0.2 mL/s (HRT of 3.16 days) and Table 2 at the flowrate of 0.67 mL/s 246 (HRT of 1.37 days). First-order reaction rate coefficients (k-rates) were calculated based on the 247 observed removals and tracer tests and compared to other results in the literature (Table 3). The 248 removal rates of the various heavy metals are presented in Fig. 4 for both HRT tested, while Fig. 249 4 shows the removal rates for nitrogen, phosphorus and phenols. 250 In general, the aerobic conditions in the wetland (aided by the aeration system) allowed relatively 251 high removal rates for Fe (96-98%), Mn (38-92%), Al (63-73%), and Zn (99%) at both HRT (Fig. 252 3), generally as oxyhydroxide precipitates. Removal of Cu (61-80%), Ni (70-85%), Pb (96-99%) 253 and Cr (63-92%) under aerobic conditions was also observed (Fig. 3), likely through co- 254 precipitation. These removal rates appear to be higher than typically reported figures for passive 255 CW systems (Aslam et al., 2007; Mustapha et al., 2018). The average pH values of the influent 256 (6.8-7.5) and effluent form all units (7.1-7.5), i.e., close to the neutral zone, also indicate that there 257 was no negative interference to the major metal removal mechanisms (i.e., adsorption, 258 precipitation, sedimentation, plant uptake) (Nyquist and Greger, 2009). The effluent pH values 259 were also within the Iranian standards for environmental discharge and effluent reuse (6-8.5; IEAP, 2602 6 261261 0 16 2008). 17 262 Table 1. Average influent (IN) and effluent (EFF) concentrations (μg/L), respective treatment 263 performance (Removal, %) and inlet mass load (g/m²/yr) of the pilot RW units at the HRT of 1.37 264 days (n=9). RW-S RW-SA RW-SP RW-SPA Inlet mass load IN EFF Rem EFF Rem EFF Rem EFF Rem (g/m²/yr) Parameter (µg/L) (µg/L) (%) (µg/L) (%) (µg/L) (%) (µg/L) (%) Al 300 110 63.3 105 65.0 100 66.7 80 73.3 2.20 As 4.37 2.43 44.4 2.15 50.8 2.25 48.5 2.16 50.6 0.032 Ce 1 0.1 90.0 0.26 74.0 0.05 95.0 0.054 94.6 0.007 Cr 6.19 1 83.8 1.52 75.4 0.48 92.2 0.62 90.0 0.045 Cu 30.58 7.76 74.6 11.95 60.9 6.21 79.7 8.0 73.8 0.224 Fe 3580 70 98.0 160 95.5 68 98.1 142 96.0 26.3 K 6510 3440 47.2 4110 36.9 1860 71.4 2080 68.0 47.7 Li 34.62 20.32 41.3 26.61 23.1 21.31 38.4 25.2 27.2 0.254 Mg 27900 22290 20.1 27050 3.0 22170 20.5 22980 17.6 204.6 Mn 130 80 38.5 30 76.9 80 38.5 25 80.8 0.953 Mo 22.29 9.33 58.1 12.83 42.4 4.62 79.3 6.58 70.5 0.163 Ni 52.96 15.95 69.9 14.5 72.6 16.01 69.8 7.87 85.1 0.388 Sc 18.53 4.32 76.7 3.73 79.9 2.9 84.3 1.8 90.3 0.136 Se 14.27 8.97 37.1 8.89 37.7 6.5 54.4 7.1 50.2 0.105 Si 10190 8630 15.3 8040 21.1 6410 37.1 7001 31.3 74.7 Sn 1.57 1.37 12.7 1.49 5.1 0.65 58.6 0.98 37.6 0.012 Sr 2480 1390 44.0 1710 31.0 1480 40.3 2025 18.3 18.2 Ta 0.25 0.13 48.0 0.17 32.0 0.11 56.0 0.16 36.0 0.002 Th 1.05 0.09 91.4 0.1 90.5 0.06 94.3 0.07 93.3 0.008 V 101 3.36 96.7 3.64 96.4 2.63 97.4 2.52 97.5 0.741 W 2.28 0.68 70.2 1.35 40.8 0.39 82.9 0.85 62.7 0.017 Zn 12820 80 99.4 180 98.6 113 99.1 128 99.0 94.0 18 Pb 62 1.78 97.1 1.82 97.1 0.8 98.7 1.1 98.2 0.455 Phenols 638 3.3 99.5 2.9 99.5 1.7 99.7 0.9 99.9 4.679 NO3-N 18000 12100 32.8 13900 22.8 8300 53.9 9100 49.4 132.0 NH4-N 31000 11300 63.5 20 99.9 20 99.9 18 99.9 227.4 PO4-P 20600 20600 0.0 20600 0.0 1000 95.1 1000 95.1 151.1 265265 266266 267 Table 2. Average influent (IN) and effluent (EFF) concentrations (μg/L), respective treatment 268 performance (Removal, %) and inlet mass load (g/m²/yr) of the pilot RW units at the HRT of 3.16 269 days (n=9). RW-S RW-SA RW-SP RW-SPA Inlet mass load IN EFF Rem EFF Rem EFF Rem EFF Rem (g/m²/yr) Parameter (µg/L) (µg/L) (%) (µg/L) (%) (µg/L) (%) (µg/L) (%) Al 300 153 49.0 120 60.0 108 64.0 91 69.7 7.38 As 4.37 3.2 26.8 2.9 33.6 3.1 29.1 0.9 79.4 0.11 Ce 1 0.12 88.0 0.13 87.0 0.05 95.0 0.05 95.0 0.02 Cr 6.19 2.3 62.8 2.5 59.6 0.48 92.2 0.62 90.0 0.15 Cu 30.58 8.8 71.2 13.4 56.2 7.01 77.1 8.8 71.2 0.75 Fe 3580 91 97.5 95 97.3 71 98.0 152 95.8 88 K 6510 2,050 68.5 4,110 36.9 1,930 70.4 760 88.3 160 Li 34.62 13.86 60.0 30.1 13.1 19.5 43.7 21.5 37.9 0.85 Mg 27900 15,390 44.8 20,015 28.3 14,510 48.0 15,310 45.1 686 Mn 130 30 76.9 10 92.3 39 70.0 10 92.3 3.20 Mo 22.29 7.37 66.9 18.4 17.5 5.93 73.4 6.61 70.3 0.55 Ni 52.96 12.18 77.0 11.63 78.0 12.52 76.4 13.71 74.1 1.30 Sc 18.53 4.29 76.8 3.89 79.0 4.2 77.3 3.5 81.1 0.46 Se 14.27 5.4 62.2 5.8 59.4 3.5 75.5 3.55 75.1 0.35 19 Si 10190 8,040 21.1 8,005 21.4 7,011 31.2 7,000 31.3 251 Sn 1.57 0.97 38.2 1.54 1.9 0.73 53.5 1.29 17.8 0.04 Sr 2480 890 64.1 2,130 14.1 980 60.5 1,560 37.1 61 Ta 0.25 0.2 20.0 0.24 4.0 0.06 76.0 0.08 68.0 0.01 Th 1.05 0.09 91.4 0.11 89.5 0.07 93.3 0.08 92.4 0.03 V 101 3.57 96.5 3.5 96.5 3.1 96.9 3.2 96.8 2.48 W 2.28 0.79 65.4 0.79 65.4 0.79 65.4 0.79 65.4 0.06 Zn 12820 60 99.5 130 99.0 78 99.4 91 99.3 315 Pb 62 2.6 95.8 2.62 95.8 1.3 97.9 1.36 97.8 1.52 Phenol 638 7.8 98.8 4.9 99.2 18 97.2 2.3 99.6 15.7 NO3-N 18000 16500 8.3 17300 3.9 1100 93.9 2400 86.7 442.4 NH4-N 31000 17300 44.2 20 99.9 6500 79.0 20 99.9 761.6 PO4-P 20600 20600 0.0 20600 0.0 900 95.6 890 95.7 506.1 270270 20 2712 7 1 Figure 4. Removal rates of heavy metals in the four pilot CW units at the two HRT tested; low 2722 7 2 21 273 (1.37 days) and high (3.16 days). 22 274 Phenols were completely removed in the units under both HRT (Fig. 4), which is in agreement 275 with results from other studies (Stefanakis et al., 2016; Schultze-Nobre et al., 2017; Gomes et al., 276 2018; Jain et al., 2020). NH4-N reached high removal in the planted units (Fig. 4) and was 277 practically completely removed in the aerated RW units apparently due to the higher dissolved 278 oxygen availability provided by plant roots and mostly by artificial aeration for the nitrification 279 process (Kadlec and Wallace, 2009; Mustapha et al., 2018; Stefanakis et al., 2019). As a result, 280 the first-order reaction rate coefficients (k; Table 4) estimated for these parameters were limited 281 by the complete removal of the mass loading, and further investigation of removal of these 282 parameters, especially phenols, in aerated CW is warranted. NO3-N removal was limited at a 283 certain level due to the enhanced aeration and conditions conducive to extensive denitrification 284 did not occur in the tested pilot configurations, although higher removal rates were measured in 285 the planted units probably due to higher plant uptake (Stefanakis et al., 2019). Phosphorus removal 286 in the system was attributable to short-term plant biomass uptake over the duration of the study 2872 period, as indicated by the high removal rates measured only in the planted units. 8 7 2882 8 8 23 2892 8 9 Figure 5. Removal rates of phenols, ammonia nitrogen, nitrate and phosphorus in the four pilot 2902 9 0 291 CW units at the two HRT tested; low (1.37 days) and high (3.16 days). 292292 293 Table 3. First-order reaction rate coefficients (k) for the racetrack wetland design and comparison 294 with available literature data. Parameter This Study Other Studies 219–290 m/yr (1.37 d HRT, range for all units) 42 m/yr (Manyin et al., 1997) Fe 82-107 m/yr (3.16 d HRT, range for 66 m/yr (Tarutis et al., 1999) all units) 169 m/yr (1.37 d HRT with Mn 21 m/yr (SF; Tarutis et al., 1999) aeration) 24 71-88 m/yr (1.37 d HRT without 10-60 m/yr (HSF with river gravel; Sikora et al., aeration) 2000) 100-600 m/yr (HSF with limestone; Sikora et al., 2000) 54-97 m/yr (1.37 d HRT, range for all units) Cu 80 m/yr (SF, estimated; Kanagy et al., 2008) 23-39 m/yr (3.16 d HRT, range for all units) 81-82 m/yr (1.37 d HRT, range for all units) Ni 15-18 m/yr (SF+HSF; Sinicrope et al., 1992) 29-42 m/yr (3.16 d HRT, range for all units) 219-284 m/yr (1.37 d HRT, range for all units) Pb 104 m/yr (SF, estimated; Kanagy et al., 2008) 95-121 m/yr (3.16 d HRT, range for all units) 52-57 m/yr 1.37 d HRT, unplanted) 148-168 m/yr (1.37 d HRT, 25-35 m/yr (SF+HSF; Sinicrope et al., 1992) planted) Cr 20-80 m/yr (SF; Kadlec and Srinivasan, 1995) 34-46 m/yr (3.16 d HRT, 3-64 m/yr (Kadlec and Wallace, 2009) unplanted) 59-66 m/yr (3.16 d HRT, planted) 28 m/yr (SF+HSF; Sinicrope et al., 1992) 360-450 m/yr (1.37 d HRT, range Zn 55-121 m/yr (SF; Gillespie et al., 2000) for all units) 21-42 m/yr (SF, estimated; Dorman et al., 2009) 25 118-146 m/yr (3.16 d HRT, range for all units) 8-17 m/yr (HSF, estimated; Rossmann et al., 2012) 107.2 m/yr (HSF; Stefanakis et al., 2016) 96 -165 m/yr (1.37 d HRT, planted 98.9 m/yr (HSF; Tatoulis et al., 2017) with aeration) Phenols 100 m/yr (SF, estimated; Polprasert and Dan, 1994) 152-201 m/yr (3.16 d HRT, planted 76 m/yr (SF, unplanted; Tang and Lu, 1993) with aeration) 150 m/yr (SF, planted; Tang and Lu, 1993) 100 m/yr (SF, estimated; Dong and Lin, 1994) 960,6 m/yr (VF, planted with aeration; Stefanakis et 233-237 m/yr (planted with al., 2019) aeration) 589 m/yr (HSF, planted with aeration; Stefanakis, 13.8 m/yr (1.37 d HRT, unplanted, 2020b) NH4-N no aeration) 2.9 m/yr (HSF, planted without aeration; Nivala et 24.4 m/yr (3.16 d HRT, unplanted, al., 2019) no aeration) 22.4 m/yr (HSF, planted without aeration; Stefanakis et al., 2016) 75 m/yr (HSF, planted with aeration; Nivala et al., 0.9-2.0 m/yr (1.37 d HRT, 2019) unplanted) 65.1 m/yr (VF, planted with aeration; Stefanakis et 51-73 m/yr (1.37 d HRT, planted) al., 2019) NO3-N 6.0-9.3 m/yr (3.16 d HRT, 96.4 m/yr (HSF, planted with aeration; Stefanakis, unplanted) 2020b) 16-19 m/yr (3.16 d HRT, planted) 3.9 m/yr (HSF, planted without aeration; Nivala et al., 2019) 26 295295 296 3.2. Removal of metals 297 As seen in Tables 1 and 2, a removal rate was detected for all measured metals, albeit at varying 298 mass loading rates and removal percentages. Generally, heavy metals removal can be attributed to 299 the (i) adsorption onto the wetland substrate (sand, gravel) and organic matter (plant roots), (ii) 300 air/water interactions, which generally lead to an aerobic (oxidizing) environment, especially in 301 the pilot units with artificial aeration, resulting in precipitation and co-precipitation of metals, and 302 (iii) uptake and assimilation in plant biomass (Sheoran and Sheoran, 2006; Kadlec and Wallace, 303 2009; García et al., 2010). 304 Iron (Fe) loading rates were 26 and 88 g/m²/yr at the two flowrates with corresponding removals 305 of 96-98% (Fig. 3). These loading rates are far below those recommended for CWs treating mine 306 drainage, which are in the range of 3,650 – 7,300 g/m²/yr (Hedin et al., 1994; PIRAMID 307 Consortium, 2003; Nyquist and Greger, 2009). The Fe effluent values of all pilot CWs were also 308 below the national limit for environmental discharge (300 μg/L; IEAP, 2008). The absence of high 309 organic loads and the respectively limited oxygen demand within the wetlands in the present study, 310 as well as the external aeration provided to the system, generally supports the hypothesis that Fe 311 was removed as an oxyhydroxide precipitate (Fe(OH)3). It is proposed that Fe removal in CWs is 312 a first-order process, with reaction rate coefficients (k) estimated at 42 m/yr (Manyin et al., 1997) 313 and 66 m/yr (Tarutis et al., 1999). Analysis of the data from this study yielded k-rates between 82- 314 107 m/yr at the longer HRT (3.16 days) and 219-290 m/yr at the shorter retention time (1.37 days) 315 (Table 3). Since virtually all Fe was removed in the CW, the higher reaction rate coefficients 316 indicate that Fe removal occurred quickly within the system. 27 317 Manganese (Mn) loads were 0.95 and 3.2 g/m²/yr at the two flowrates (Tables 1 and 2), which is 318 considerably lower than the 182.5 g/m²/yr recommended for CWs treating mine drainage 319 (PIRAMID Consortium, 2003). Mn removal was consistently higher in the CW units with artificial 320 aeration at both flowrates. At the HRT of 1.37 days, Mn removal was 70-77% without aeration 321 and 92% with aeration (Fig. 3); for the HRT of 3.16 days, Mn removal was 38.5% without aeration 322 and 77-81% with aeration (Fig. 3). As the kinetics for Mn oxidation and precipitation are generally 323 slower compared to Fe (Hedin and Nairn, 1993) and Mn removal is not significant in aerobic CWs 324 with low Fe concentration (Hallberg and Johnson, 2005), the enhancement of the aerobic 325 conditions in the aerated wetlands could potentially have aided in Mn removal. The Mn effluent 326 values of all pilot CWs were also below the national limit for environmental discharge (50 μg/L; 327 IEAP, 2008). Mn removal as a first-order process in CWs has been proposed. Data analysis 328 indicated reaction rate coefficients (k) of 169 m/yr for the optimum conditions (highest loading 329 with aeration), and 71-88 m/yr (highest loading without aeration) (Table 3). In the literature, 21 330 m/yr has been proposed for surface flow (SF) CW (Tarutis et al., 1999). For HSFCWs (without 331 aeration), rates of 10-60 m/yr for river gravel and 100-600 m/yr for limestone-based systems have 332 been estimated (Sikora et al., 2000). 333 Aluminum (Al) can form hydroxide precipitates under oxidizing conditions, which are strongly 334 associated with phosphorus retention. Al loads were 2.2 and 7.4 g/m²/yr (Tables 1 and 2), 335 compared to loads of approximately 0.44 g/m²/yr observed for municipal wastewater applications 336 of non-aerated HSFCWs (Vymazal and Krása, 2003). The removal percentages in this study 337 ranged from 63-73% at the HRT of 3.16 days, and 49-70% at the HRT of 1.37 days (Fig. 3), 338 indicating that longer HRT resulted in better Al removal. Respective figures for Al reported in the 339 literature are approximately 70% (Gensemer and Playle, 1999), and are comparable to this study. 28 340 Copper (Cu) loads were 0.75 and 0.22 g/m²/yr at the two flowrates (Tables 1 and 2), with removal 341 percentages between 56-80%. While Cu removal is often associated with reducing conditions and 342 sulfide precipitation, removal under aerobic conditions via co-precipitation with Fe and Mn 343 oxyhydroxide precipitates is also possible. For mining applications, loading rates of 3,650 g/m²/yr 344 (sulfide producing conditions) and 18.25 g/m²/yr (aerobic conditions) have been proposed 345 (PIRAMID Consortium, 2003), which are considerably higher than this study. The Cu effluent 346 values of all pilot CWs were below the national limit for environmental discharge (20 μg/L; IEAP, 347 2008). Cu removal as a first-order process under aerobic conditions is a possibility (Tarutis et al., 348 1999). If Cu removal is through co-precipitation with Fe and Mn, and those metals follow a first- 349 order process, reaction rate coefficients (k) can also be estimated for Cu. Data analysis indicates 350 k-rates of 23-39 m/yr at the lower HRT, and 54-97 m/yr at the higher HRT (Table 3). In a pilot- 351 scale SFCW treating oil produced water, Cu removal data indicated a k-rate of 80 m/yr (Kanagy 352 et al., 2008). 353 Nickel (Ni) loads were 1.30 and 0.39 g/m²/d at the two flowrates, with removal percentages 354 between 70-85% (HRT of 3.16 days) and 74-77% (HRT of 1.37 days) (Tables 1 and 2). While Ni 355 removal is often associated with reducing conditions and sulfide precipitation, removal under 356 aerobic conditions through co-precipitation with Fe and Mn oxyhydroxide precipitates is also 357 possible. For mining applications, loading rates of 730 g/m²/yr (sulfide producing conditions) and 358 14.6 g/m²/yr (aerobic conditions) have been proposed (PIRAMID Consortium, 2003), which are 359 considerably higher than this study. Ni removal as a first-order process under aerobic conditions 360 is a possibility (Tarutis et al., 1999), but data sets to estimate a first-order reaction rate coefficient 361 (k) are limited. Data from a SF + HSF CW at Sacramento, California resulted in estimated k-rates 362 of 15-18 m/yr (Sinicrope et al., 1992), at Ni loading rates of 5.5-6.7 g/m²/yr (Kadlec and Wallace, 29 363 2009). Data from this study yields k-rate estimates of 29-48 m/yr (HRT of 3.16 days) and 81-82 364 m/yr (HRT of 1.37 days) (Table 3). 365 Lead (Pb) loads were 1.52 and 0.46 g/m²/yr at the two flowrates (Tables 1 and 2). Almost all Pb 366 was removed (96-99%; Fig. 3), which is consistent with the formation of very insoluble sulfate 367 and carbonate precipitates (DeVolder et al., 2003). The Pb effluent values of all pilot CWs were 368 below the national limit (50 μg/L; IEAP, 2008). Pb removal as a first-order process has been 369 proposed, but previous data sets did not allow for the estimation of a k-rate (Kadlec and Wallace, 370 2009). Data from this study yields k-rate estimates of 95-121 m/yr at the lower HRT and 219-284 371 m/yr at the higher HRT (Table 3). Since virtually all Pb was removed at both HRTs, the use of the 372 higher k-rate estimates appears justified. In a pilot-scale SFCW treating oil produced water, Pb 373 removal data indicated a k-rate of approximately 104 m/yr (Kanagy et al., 2008). 374 Chromium (Cr) loads were 0.15 and 0.05 g/m²/yr at the two flowrates (Tables 1 and 2). Cr removal 375 was affected by plants’ presence. The removal in the unplanted units were 75-84% (low loading 376 rate) and 60-63% (high loading rate), and 90-92% in the planted units at both loading rates (Fig. 377 3). On the other hand, the artificial aeration did not appear to influence Cr removal. The Cr influent 378 and effluent values of all pilot CWs were below the national limit for environmental discharge (50 379 μg/L; IEAP, 2008). Cr can be removed through co-precipitation with Fe and Mn oxyhydroxide 380 precipitates; the first-order modeling approach applied to those metals can also be applied to Cr. 381 A summary of Cr removal data estimated k-rate coefficients ranging from 3-64 m/yr, with a median 382 value of 19 m/yr (Kadlec and Wallace, 2009). Data from SFCW and HSFCW mesocosms at 383 Sacramento, California treating municipal wastewater yielded k-rate estimates of 25-35 m/yr 384 (Sinicrope et al., 1992), and data from a SFCW mesocosm treating petroleum wastewater yielded 385 k-rate estimates of 20-80 m/yr (Kadlec and Srinivasan, 1995). In a pilot-scale SFCW treating ash 30 386 basin water, Cr removal data suggested k-rates of 29-36 m/yr (Dorman et al., 2009). In this study, 387 the unplanted units resulted in k-rate estimates of 34-46 m/yr (HRT of 3.16 days) and 52-57 m/yr 388 (HRT of 1.37 days) and the planted units in 59-66 m/yr (HRT of 3.16 days) and 148-168 m/yr 389 (HRT of 1.37 days) (Table 3). 390 Zinc (Zn) loads were 315 and 94 g/m²/yr at the two flowrates (Tables 2 and 3). Zn was almost 391 completely removed (99%; Fig. 3), which is consistent with the formation of insoluble carbonate 392 precipitates, and through co-precipitation with Fe and Mn oxyhydroxide precipitates (Younger et 393 al., 2002; Kadlec and Wallace, 2009). In comparison, recommendations for loading rates for mine 394 water treatment range from 14.6 g/m²/yr (PIRAMID Consortium, 2003) up to 2,555 g/m²/yr 395 (Younger et al., 2002). The Zn effluent values of all pilot CWs were below the national limit for 396 environmental discharge (500 μg/L; IEAP, 2008). Zn removal can be considered a first-order 397 process. For a SFCW and HSFCW mesocosm at Sacramento, California, the estimated k-rate was 398 28 m/yr (Sinicrope et al., 1992), while for a SFCW treating oil refinery effluent, the k-rate for Zn 399 removal was estimated at 55-121 m/yr at two different loading rates (Gillespie et al., 2000; Kadlec 400 and Wallace, 2009). In a pilot-scale CW treating oil produced water (Kanagy et al., 2008), Zn 401 removal data indicated a k-rate of approximately 15 m/yr. In another pilot-scale SFCW treating 402 ash basin water, Zn removal data suggested k-rates of 21-42 m/yr (Dorman et al., 2009). In this 403 study, the estimated k-rates ranged from 118-146 m/yr (low HRT) to 360-450 m/yr (high HRT) 4044 (Table 3). 0 4 4054 0 5 406 3.3. Phenols 407 Phenol was essentially completely removed in all systems (Fig. 4), as the influent (638 μg/L) was 408 removed to close to non-detect concentrations (0.9-18 μg/L; Tables 1 and 2). First order rate 31 409 coefficients (k) ranged from 152 – 201 m/yr (HRT of 3.16 days) and 96 – 165 m/yr (HRT of 1.37 410 days). It is reported that aerobic biodegradation is the main phenol removal mechanism in CW 411 (Kurzbaum et al., 2010; Rossmann et al., 2012; Stefanakis et al., 2016), though anaerobic 412 degradation is also possible (Kumaran and Paruchuri, 1997). 413 Similarly, high removal rates for phenols are reported in CW studies in the literature, even for 414 higher influent concentrations. Data from a series of pilot-scale HSFCW treating coffee processing 415 wastewater in Brazil (Rossmann et al., 2012) suggests phenol degradation k-rates of 8-17 m/yr. A 416 pilot HSFCW treating cork boiling wastewater in Portugal suggests k-rates between 35-42 m/yr 417 (Gomes et al., 2018). In a series of pilot studies using SFCW (Polprasert et al., 1996), data 418 indicated a phenol degradation k-rate of approximately 100 m/yr. In a study of planted and 419 unplanted SFCW channels in China, k-rates for phenol degradation were estimated at 150 m/yr 420 (planted) and 76 m/yr (unplanted) (Tang et al., 1993). In a study at a petroleum facility, a full-scale 421 SFCW (50 ha) had an estimated k-rate of approximately 100 m/yr for phenol removal (Dong and 422 Lin, 1994). Complete phenols removal and a k rate of 107.2 m/yr was found in a HSFCW system 423 treating groundwater contaminated with phenols and petroleum derivatives (Stefanakis et al., 424 2016). A similar phenols removal k rate of 98.9 m/yr is reported for a HSFCW treating pre-treated 4254 olive mill wastewater (Tatoulis et al., 2017). 2 5 4264 2 6 427 3.4. Nitrogen 428 Ammonia nitrogen (NH4-N) removal (nitrification) occurred in all pilot units at both loading rates 429 (influent of 31,000 μg/L, effluent of 18-20 μg/L; Tables 1 and 2), with corresponding k-rates of 430 233-237 m/yr (Table 3). NH4-N removal was strongly influenced by the presence of aeration (Fig. 431 4). It is likely the actual k-rates are higher, since the systems ran out of NH4-N with essentially 32 432 complete removal. In a similar study, an aerated HSFCW in Germany had an estimated k-rate of 433 1,540 m/yr vs. 2.9 m/yr for a non-aerated system (Table 3; Nivala et al., 2019). A high k-rate of 434 960.6 m/yr was also estimated for an aerated VFCW in the UK operating as tertiary treatment stage 435 (Stefanakis et al., 2019), while an aerated second-stage HSFCW in Oman suggests k-rate of 589 436 m/yr (Stefanakis, 2020b). In both these systems under different climates, complete ammonia 437 removal occurred as well, indicating the positive impact of artificial aeration on aerobic 438 nitrification (Stefanakis et al., 2019). 439 Ammonia removal was also influenced by the presence of vegetation (Tables 1 and 2), since the 440 presence of plants promotes the biofilm development and the related aerobic oxidation of NH4-N 441 by microorganisms (Stefanakis, 2016). Many studies report that plants’ presence favor ammonia 442 removal (Kadlec and Wallace, 2009; Stefanakis et al., 2016; Schultze-Nobre et al., 2017) The 443 shorter duration trial (HRT of 1.37 days) had an estimated k-rate of 39 m/yr, whereas at the longer 444 HRT (3.16 days), the estimated k-rate was 233 m/yr (since the system ran out of NH4-N). Without 445 plants or aeration, k-rates were lower, 13.8 m/yr (HRT of 1.37 days) and 24.4 m/yr (HRT of 3.16 446 days), respectively. Respective k-rate of 22.4 and 39.9 m/yr are reported for HSF systems treating 447 contaminated groundwater and cork boiling wastewater (Stefanakis et al., 2016; Gomes et al., 448 2018). 449 Nitrate (NO3-N) removal (denitrification) also occurred but al lower levels. Nitrate concentrations 450 decreased from influent values, despite the conversion of NH4-N to oxidized forms of nitrogen 451 within the pilot systems. Aeration possibly impacted NO3-N removal, as more NH4-N was 452 converted to oxidized forms of nitrogen, and the aerobic conditions were not conducive to 453 complete denitrification. For systems without vegetation, estimated k-rates with and without 454 aeration were 0.9-2.0 m/yr (HRT of 1.37 days) and 6.0-9.3 m/yr (HRT of 3.16 days), respectively. 33 455 With vegetation, NO3-N removal rates increased (Fig. 4), potentially due to plant uptake 456 considering that ammonia is completely oxidized and nitrate remain the only available nitrogen 457 form for plant uptake, as well as due to the potential presence of more bioavailable carbon in the 458 system, but again, NO3-N removal rates were possibly by aeration. However, the presence of non- 459 aerated sections in the design possibly limited the effect of the aeration in the denitrification extent 460 in the system. The fact that aeration also enhances the microbial growth (Chazarenc et al., 2009; 461 Nivala et al., 2020) indicates that nitrate removal should be a combination of microbial oxidation 462 and plant uptake. Estimated k-rates with and without aeration were 51-73 m/yr (HRT of 1.37 days) 463 and 16-19 m/yr (HRT of 3.16 days). In a similar study, an aerated HSFCW in Germany (Nivala et 464 al., 2019) had an estimated k-rate of 75 m/yr with aeration and 3.9 m/yr without aeration, since 465 NO3-N was not produced in the non-aerated wetland. Similar k-rates of 65.1 and 96.4 m/yr were 466 also estimated for an aerated VFCW in the UK (Stefanakis et al., 2019) and an aerated HSFCW in 467 Oman (Stefanakis, 2020b), with the latter receiving small amounts of raw wastewater (step- 4684 feeding) to increase the carbon availability, hence the higher rate observed. 6 8 4694 6 9 470 3.5. Phosphorus 471 Phosphorus (PO4-P) removal only occurred when vegetation was present in the system, since it is 472 known that PO4-P removal in CW systems is generally lower compared to other pollutants and 473 mostly depends on the adsorption capacity of the substrate media and the uptake by plants to cover 474 their growth needs (Kadlec and Wallace, 2009; García et al., 2010; Stefanakis, 2016). For the trials 475 conducted with vegetation, phosphorus uptake was essentially complete (95-98%) with effluent 476 concentrations close to detection limits. The applied phosphorus loads of 151 and 507 g/m²/yr (at 34 477 the low and high loading rates) greatly exceeds the amount of phosphorus required to support the 35 478 plant biomass cycle (Kadlec and Wallace, 2009). As a result, it is anticipated that the system would 479 saturate with phosphorus in a longer-term study as the plant community matures and the media 4804 gets gradually saturated. 8 0 4814 8 1 482 3.6. Optimum design configuration 483 The tested design proved to be efficient in treating the refinery effluent reaching higher removal 484 rates for most of the tested parameters. The different pilot units revealed the significant role of 485 plants in the overall performance, as the planted bed (RW-SP) showed better removal rates for all 486 parameters than the unplanted unit (RW-S) (Tables 2 and 3), with statistically significant removals 487 (p<0.05) occurring for nitrogen and phosphorus and many heavy metals (Al, Ce, Cr, K, Mo, Sc, 488 Se, Si, Sn, Sr, Ta, W). The performance appears to be further enhanced with the addition of the 489 aerated section (RW-SPA; Tables 2 and 3). Artificial aeration seems to improve (p<0.05) the 490 removal of ammonia nitrogen, Al, As, K, Mn, Sc, and Se compared to the planted non-aerated unit 491 (RW-SP). On the other hand, when the aeration was applied in the unplanted unit (RW-SA), it did 492 not improve the overall performance. Among the two HRT tested, high removals were observed 493 even for the smaller HRT of 1.4 days with slightly higher removal rates compared to the removals 494 under the HRT of 3.2 days and with statistical significance (p<0.05) occurring for nitrate, Al, As, 495 K, Li, Mg, Mn, Ni, Se, Sr, and Ta. 496 The advantage of the tested design arises from the comparison with other wetland designs applied 497 for similar effluent treatment. The surface flow CW is widely applied for such effluents, but it has 498 a significantly higher area demand due to the prolonged HRT required, e.g., 20 days under similar 499 hydraulic load of 0.06 m/d (Mustafa et al., 2018). A VFCW operating at 0.07 m/d and a HRT of 2 36 500 days showed a lower performance for nutrients though receiving a much lower influent 37 501 concentration since it received a secondary refinery effluent (Mustapha et al., 2015; 2018). A 502 HSFCW was also tested for diesel contaminated water at different levels, resulting in optimum 503 removal conditions of 63 days retention time (Al-Baldawi et al., 2014). In a review paper on 504 refinery effluent treatment in CW, the HRT range reported varies between 2.1-90 days, with the 505 majority of the systems being in the range above 10 days (Jia et al., 2020). In the same review 506 study, the positive role of plants and aeration is also indicated, while the HSFCW is reported to be 5075 the design with the better performance. 0 7 5085 0 8 509 4. Conclusions 510 Petroleum refinery effluents represent a major industrial pollution source of significant concern 511 due to its toxic and harmful nature. Established biological methods are not very effective in the 512 remediation of these effluents, while advanced oxidation processes, although effective, are 513 expensive to install and operate. In this study, the technology of Constructed Wetlands was applied 514 for refinery effluent containing heavy metals, phenols and nitrogen as a sustainable nature-based 515 treatment solution. An innovative CW design forming a meandric ‘racetrack’ shape and combining 516 different treatment environments (aerobic and anaerobic) was tested for the first time at pilot-scale. 517 Different construction (non-aerated, artificial aeration, planted and unplanted) and operation 518 (different hydraulic loading rates) parameters were studied to determine the optimum design for 519 effective performance. The first-order removal of metals and the simultaneous removal of phenols 520 and nutrients from the oil refinery wastewater were calculated. The planted unit with artificial 521 aeration managed to reach high removal rates for Fe (96-98%), Mn (38-81%), Al (49-73%), Zn 522 (99-100%), Cu (61-80%), Ni (70-85%), Pb (96-99%) and Cr (60-92%) even at the short HRT of 38 523 1.3 days, while complete removal of phenols and full nitrification was also found. As the calculated 39 524 first-order removal coefficients showed, the overall performance of this novel design exceeded the 525 reported efficiency of the common CW designs and implies the potential to further apply and test 526 it for other industrial effluents. 527 528 Authorship contribution statement 529 Mozaffari M.S: Conceptualization, Writing - review, Visualization, Investigation, Data curation. 530 Shafiepour E.: Project administration. Mirbagheri S.A.: Resources, Funding acquisition 531 Rakhshandehroo G.: Methodology, Funding acquisition. Wallace S.: Conceptualization, Formal 532 analysis, Data curation. Stefanakis A.I.: Formal analysis, Data curation and analysis, Writing - 5335 original draft, Writing - review 3 3 5345 3 4 535 Acknowledgements 536 This study work was supported by the Civil Engineering Department, Shiraz University and the 5375 Khajeh Nasir Toosi University of Technology. 3 7 5385 3 8 539 References 540 APHA, 1999. Standard Methods for the Examination of Water and Wastewater Part. American 541 Water Works Association (AWWA). 542 Al-Baldawi, I.A.W., Abdullah, S.R.S., Hasan, H.A.S., Suja, F., Anuar, N., Mushrifah, I., 2014. 543 Optimized conditions for phytoremediation of diesel by Scirpus grossus in horizontal 40 544 subsurface flow constructed wetlands (HSFCWs) using response surface methodology. J. 545 Environ. Manage. 140, 152-159. 41 546 Aslam, M.M., Mali, M., Baig, M.A., Qazi, I.A., Iqbal, J., 2007. Treatment performances of 547 compost-based and gravel-based vertical flow wetlands operated identically for refinery 548 wastewater treatment in Pakistan. Ecol. Eng. 30, 34-42, 549 https://doi.org/10.1016/j.ecoleng.2007.01.002 550 Bidhendi, N.G.R., Mehrdadi, N., Mohammadnejad, S., 2010. Water and wastewater minimization 551 in Tehran oil refinery using water pinch analysis. Int. J. Environ. Res. 4(4), 583-594, 552 doi:10.22059/IJER.2010.244 553 Boog, J., 2013. Effect of the aeration scheme on the treatment performance of intensified treatment 554 wetland systems. TU Bergakademie Freiberg: Freiberg, Germany. 555 Boog, J., Nivala, J., Aubron, T., Wallace, S., van Afferden, M., Müller, R.A., 2014. Hydraulic 556 characterization and optimization of total nitrogen removal in an aerated vertical subsurface 557 flow treatment wetland. Bioresour. Technol. 162, 166-74, 558 https://doi.org/10.1016/j.biortech.2014.03.100 559 Chazarenc, F.; Gagnon, V.; Comeau, Y.; Brisson, J. Effect of plant and artificial aeration on solids 560 accumulation and biological activities in constructed wetlands. Ecol. Eng. 2009, 35, 1005– 561 1010. 562 Del Rio, L., Hadwin, A.K.M., Pinto, L.J., MacKinnon, M.D., Moore, M.M., 2006. Degradation of 563 naphthenic acids by sediment micro‐ organisms. J. Appl. Microbiol. 101(5), 1049-1061, 564 https://doi.org/10.1111/j.1365-2672.2006.03005.x 565 DeVolder, P.S., Brown, S.L., Hesterberg, D., Pandya, K., 2003. Metal bioavailability and 566 speciation in a wetland tailings repository amended with biosolids compost, wood ash, and 567 sulfate. J. Environ. Qual. 32(3), 851-864, https://doi.org/10.2134/jeq2003.8510 42 568 Dong, K., Lin, C., 1994. Treatment of petrochemical wastewater by the system of wetlands and 569 oxidation ponds and engineering design of the system. Guangzhou, P.R. China: IWA. 570 Dorman, L., Castle, J.W., Rodgers, Jr., J.H., 2009. Performance of a pilot-scale constructed 571 wetland system for treating simulated ash basin water. Chemosphere 75(7), 939-47, 572 https://doi.org/10.1016/j.chemosphere.2009.01.012 573 García, J., Rousseau, D.P.L., Morató, J., Lesage, E., Matamoros, V., Bayona, J.M., 2010. 574 Contaminant removal processes in subsurface-flow constructed wetlands: a review. Crit. Rev. 575 Environ. Sci. Technol. 40, 561-661, https://doi.org/10.1016/j.chemosphere.2009.01.012 576 Gensemer, R.W., Playle, R.C., 1999. The bioavailability and toxicity of aluminum in aquatic 577 environments. Crit. Rev. Environ. Sci. Technol. 29(4), 315-450, 578 https://doi.org/10.1080/10643389991259245 579 Gholipour, A., Zahabi, H., Stefanakis, A.I., 2020. A novel pilot and full-scale constructed wetland 580 study for glass industry wastewater treatment. Chemosphere 247, 125966, 581 https://doi.org/10.1016/j.chemosphere.2020.125966 582 Gillespie Jr, W.B., Hawkins, W.B., Rodgers Jr, J.H., Cano, M.L., Dorn, P.B., 2000. Transfers and 583 transformations of zinc in constructed wetlands: Mitigation of a refinery effluent. Ecol. Eng. 584 14(3), 279-292, https://doi.org/10.1006/eesa.1999.1762 585 Gomes, A.C., Silva, L., Albuquerque, A., Simões, R., Stefanakis, A.I., 2018. Investigation of lab- 586 scale horizontal subsurface flow constructed wetlands treating industrial cork boiling 587 wastewater. Chemosphere 207, 430-439, https://doi.org/10.1016/j.chemosphere.2018.05.123 588 Gorito, A.M., Ribeiro, A.R., Almeida, C.M.R., Silva, A.M.T., 2017. A review on the application 589 of constructed wetlands for the removal of priority substances and contaminants of emerging 43 590 concern listed in recently launched EU legislation. Environ. Pollut. 227, 428-443, 591 https://doi.org/10.1016/j.envpol.2017.04.060 592 Hallberg, K.B., Johnson, D.B., 2005. Biological manganese removal from acid mine drainage in 593 constructed wetlands and prototype bioreactors. Sci. Total Environ. 338, 115-124, 594 https://doi.org/10.1016/j.scitotenv.2004.09.011 595 Hedin, R.S, Nairn, R.W., 1993. Contaminant removal capabilities of wetlands constructed to treat 596 coal mine drainage. In: Constructed Wetlands for Water Quality Improvement, G.A. Moshiri, 597 Editor. CRC Press: Boca Raton, Florida, p. 187-196. 598 Hedin, R.S., Nairn, R.W., Kleinman, R.L.P., 1994. Passive treatment of coal mine drainage with 599 limestone. U.S. Bureau of Mines: Pittsburgh, Pennsylvania, p. 35. 600 IEPA, 2008. Environmental regulations and standards of Iran. Iran Global Environmental Law. 601 Tehran, pp. 234-39. 602 Jain, M., Majumder, A., Ghosal, P.S., Gupta, A.K., 2020. A review on treatment of petroleum 603 refinery and petrochemical plant wastewater: A special emphasis on constructed wetlands. J. 604 Environ. Manage. 272, 111057, https://doi.org/10.1016/j.jenvman.2020.111057 605 Ji, G.D., Sun, T.H., Ni, J.R., 2007. Surface flow constructed wetland for heavy oil-produced water 606 treatment. Bioresour. Technol. 98, 436–441, https://doi.org/10.1016/j.biortech.2006.01.017 607 Johnson, B.M., Kanagy, L.E., Rodgers Jr, J.H., Castle, J.W., 2008. Feasibility of a pilot-scale 608 hybrid constructed wetland treatment system for simulated natural gas storage produced waters. 609 Environ. Geosci. 15(3), 91-104, https://doi.org/10.1306/eg.06220707004 610 Kadlec, R.H., Wallace S.D., 2009. Treatment Wetlands, Second Edition. Boca Raton, Florida: 611 CRC Press. 44 612 Kadlec, R.H., Srinivasan, K., 1995. Wetland Treatment of Oil and Gas Well Wastewaters. U.S. 613 Department of Energy: Bartlesville, Oklahoma. 614 Kanagy, L., Johnson, B.M., Castle, J.W., Rodgers Jr, J.H., 2008. Design and performance of a 615 pilot-scale constructed wetland treatment system for natural gas storage produced water. 616 Bioresour. Technol. 99(6). 1877-1885, https://doi.org/10.1016/j.biortech.2007.03.059 617 Kumaran, P., Paruchuri, Y.L., 1997. Kinetics of phenol biotransformation. Water Res. 31(1), 11- 618 22, https://doi.org/10.1016/S0043-1354(99)80001-3 619 Kurzbaum, E., Kirzhner, F., Sela, S., Zimmels, Y., Armon, R., 2010. Efficiency of phenol 620 biodegradation by planktonic Pseudomonas pseudoalcaligenes (a constructed wetland isolate) 621 vs. root and gravel biofilm. Water Res, 44(17), 5021-5031. 622 Langwaldt, J., Puhakka, J., 2000. On-site biological remediation of contaminated groundwater: a 623 review. Environ. Pollut. 2000. 107(2), 187-197, https://doi.org/10.1016/S0269-7491(99)00137- 624 2 625 Manyin, T., Williams, F.M., Stark, L.R., 1997. Effects of iron concentration and flow rate on 626 treatment of coal mine drainage in wetland mesocosms: An experimental approach to sizing of 627 constructed wetlands. Ecol. Eng. 9(3-4), 171-186, https://doi.org/10.1016/S0925- 628 8574(97)10005-2 629 Mustafa, A., Azim, M.K., Raza, Z., Kori, J.A., 2018. BTEX removal in a modified free water 630 surface wetland. Chem. Eng. J. 333, 451–455. 631 Mustapha, H.I., van Bruggen, J.J.A., Lens, P.N.L., 2015. Vertical subsurface flow constructed 632 wetlands for polishing secondary Kaduna refinery wastewater in Nigeria. Ecological 633 Engineering 84, 588–595. 45 634 Mustapha, H.I., van Bruggen, J.J.A., Lens, P.N.L., 2018. Fate of heavy metals in vertical 635 subsurface flow constructed wetlands treating secondary treated petroleum refinery wastewater 636 in Kaduna, Nigeria. Int. J. Phytoremed. 20(1), 44-53, 637 https://doi.org/10.1080/15226514.2017.1337062 638 Nivala, J., Kahl, S., Boog, J., van Afferden, M., Reemtsma, T., Müller, R., 2019. Dynamics of 639 emerging organic contaminant removal in conventional and intensified subsurface flow 640 treatment wetlands. Sci. Total Environ. 649, 1144-1156, 641 https://doi.org/10.1016/j.scitotenv.2018.08.339 642 Nivala J., Murphy, C., Freeman, A., 2020. Recent Advances in the Application, Design, and 643 Operations & Maintenance of Aerated Treatment Wetlands. Water 12(4), 1188; 644 https://doi.org/10.3390/w12041188 645 Nyquist, J., Greger, M., 2009. A field study of constructed wetlands for preventing and treating 646 acid mine drainage. Ecol. Eng. 35(5), 630-642, https://doi.org/10.1016/j.ecoleng.2008.10.018 647 PIRAMID Consortium, 2003. Engineering guidelines for the passive remediation of acidic and/or 648 metalliferous mine drainage and similar wastewaters. European Commission 5th Framework 649 RTD Project no. EVK1-CT-1999-000021, Passive in-situ remediation of acidic mine-industrial 650 drainage, University of Newcastle Upon Tyne: Newcastle Upon Tyne, UK. p.166. 651 Polprasert, C., Dan, N.P., Thayalakumaran, N., 1996. Application of constructed wetlands to treat 652 some toxic wastewaters under tropical conditions. Water Sci. Technol. 34(11), 165-171, 653 https://doi.org/10.1016/S0273-1223(96)00834-7. 654 Qureshi, N., Annous, B.A., Ezeji, T.C., Karcher, P., Maddox, I.S., 2005. Biofilm reactors for 655 industrial bioconversion processes: employing potential of enhanced reaction rates. Microbial 656 Cell Factories 4, 24, https://doi.org/10.1186/1475-2859-4-24. 46 657 Ramírez, S., Torrealba, G., Lameda-Cuicas, E., Molina-Quintero, L., Stefanakis, A.I., 2019. 658 Investigation of pilot-scale Constructed Wetlands treating simulated pre-treated tannery 659 wastewater under tropical climate. Chemosphere 234, 496-504, 660 https://doi.org/10.1016/j.chemosphere.2019.06.081 661 Ramond, J-B., Welz, P.J., Cowan, D.A., Burton, S.G., 2012. Microbial community structure 662 stability, a key parameter in monitoring the development of constructed wetland mesocosms 663 during start-up. Research in Microbiology 163(1), 28-35. 664 Rossmann, M., de Matos, A.T., Abreu, E.C., Silva, F.F., Borges, A.C., 2012. Performance of 665 constructed wetlands in the treatment of aerated coffee processing wastewater: Removal of 666 nutrients and phenolic compounds. Ecol. Eng. 2012. 49, 264-269, 667 https://doi.org/10.1016/j.ecoleng.2012.08.017 668 Schultze-Nobre, L., Wiessner, A., Bartsch, C., Paschke, H., Stefanakis, A.I., Aylward, L.A., 669 Kuschk, P., 2017. Removal of dimethylphenols and ammonium in laboratory-scale horizontal 670 subsurface flow Constructed Wetlands. Eng. Life Sci. 17(12), 1224-1233, 671 https://doi.org/10.1002/elsc.201600213 672 Sheoran, A.S., Sheoran, V., 2006. Heavy metal removal mechanism of acid mine drainage in 673 wetlands: A critical review. Mineral Eng. 19, 105-116, 674 https://doi.org/10.1016/j.mineng.2005.08.006 675 Sikora, F.J., Behrends, L.L., Brodie, G.A., Taylor, H.N., 2000. Design criteria and required 676 chemistry for removing manganese in acid mine drainage using subsurface flow wetlands. 677 Water Environ. Res. 72(5), 536-544, https://doi.org/10.2175/106143000X138111 47 678 Sinicrope, T.L., Langis, R., Gersberg, R.M., Busnardo, M.J., Zedler, J.B., 1992. Metal removal by 679 wetlands mesocosms subjected to different hydroperiods. Ecol. Eng. 1(4), 309-322, 680 https://doi.org/10.1016/0925-8574(92)90013-R 681 Stefanakis, A.I., 2016. Constructed Wetlands: description and benefits of an eco-tech water 682 treatment system. In: McKeown, A.E., Bugyi, G. (Eds.), Impact of Water Pollution on Human 683 Health and Environmental Sustainability. IGI Global, Hershey PA, USA, pp. 281-303, 684 doi:10.4018/978-1-4666-9559-7.ch012 685 Stefanakis, A.I., 2018. Constructed wetlands for industrial wastewater treatment. John Wiley & 686 Sons Ltd, Chichester, UK, doi:10.1002/9781119268376 687 Stefanakis, A.I., 2019. The role of constructed wetlands as green infrastructure for sustainable 688 urban water management. Sustainability 11(24), 6981, https://doi.org/10.3390/su11246981 689 Stefanakis, A.I., 2020a. The fate of MTBE and BTEX in Constructed Wetlands. Appl. Sci. 10, 690 127; doi:10.3390/app10010127. 691 Stefanakis, A.I., 2020b. Constructed wetlands for sustainable wastewater treatment in hot and arid 692 climates: opportunities, challenges and case studies in the Middle East. Water 12(6), 1665, 693 https://doi.org/10.3390/w12061665 694 Stefanakis, A.I., Seeger, E., Dorer, C., Sinke, A., Thullner, M., 2016. Performance of pilot-scale 695 horizontal subsurface flow constructed wetlands treating groundwater contaminated with 696 phenols and petroleum derivatives. Ecol. Eng. 95, 514-526, 697 https://doi.org/10.1016/j.ecoleng.2016.06.105 698 Stefanakis, A.I., Prigent, S., Breuer, R., 2018. Integrated produced water management in a desert 699 oilfield using wetland technology and innovative reuse practices. in: Stefanakis, A.I. (ed.), 48 700 Constructed wetlands for industrial wastewater treatment. John Wiley & Sons Ltd, Chichester, 701 UK, pp. 25-42, https://doi.org/10.1002/9781119268376.ch1 702 Stefanakis, A.I., Bardiau, M., Silva, D., Taylor, H., 2019. Presence of bacteria and bacteriophages 703 in full-scale trickling filters and an aerated constructed wetland. Sci. Total Environ. 659, 1135– 704 1145, https://doi.org/10.1016/j.scitotenv.2018.12.415 705 Tang, S.Y., Lu, X.W., 1993. The use of Eichhornia crassipes to cleanse oil refinery wastewater in 706 China. Ecol. Eng. 2(3), p. 243-251, https://doi.org/10.1016/0925-8574(93)90017-A 707 Tarutis, W.J., Stark, L.R., Williams, F.M., 1999. Sizing and performance estimation of coal mine 708 drainage wetlands. Ecol. Eng. 12(4), 353-372, https://doi.org/10.1016/S0925-8574(98)00114- 709 1 710 Tatoulis, T., Stefanakis, A.I., Frontistis, Z., Akratos, C.S., Tekerlekopoulou, A.G., Mantzavinos, 711 D., Vayenas, D.V., 2016. Treatment of table olive washing water using trickling filters, 712 constructed wetlands and electrooxidation. Environ. Sci. Pollut. Res. 1-8, doi:10.1007/s11356- 713 016-7058-6. 714 Vymazal, J., Krasa, P., 2003. Distribution of Mn, Al, Cu and Zn in a constructed wetland receiving 715 municipal sewage. Water Sci. Technol. 48(5), 299-305, https://doi.org/10.2166/wst.2003.0336 716 Wallace, S.D., 2002. Method for removing pollutants from water. United States. 717 Wallace, S., Schmidt, M., Larson, E., 2011. Long term hydrocarbon removal using treatment 718 wetlands. SPE Annual Technical Conference and Exhibition, Denver, Colorado, USA, SPE- 719 145797-MS, https://doi.org/10.2118/145797-MS 720 Wu, S., Wallace, S., Brix, H., Kuschk, P., Kirui, W.K., Masi, F., Dong, R., 2015. Treatment of 721 industrial effluents in constructed wetlands: Challenges, operational strategies and overall 722 performance. Environ. Pollut. 201, 07-120, https://doi.org/10.1016/j.envpol.2015.03.006 49 723 Younger, P.L., Banwart, S.A., Hedin, R.S., 2002. Mine Water: Hydrology, Pollution, 724 Remediation. Kluwer Academic Publishers: London, United Kingdom. doi:10.1007/978-94- 725 010-0610-1 726 Xin, J., Tang, J., Liu, Y., Zhang, Y., Tian, R., 2019. Pre-aeration of the rhizosphere offers potential 727 for phytoremediation of heavy metal-contaminated wetlands. J. Hazard. Mater. 374, 437-446, 728 https://doi.org/10.1016/j.jhazmat.2019.04.010 50 Table Click here to access/download;Table;List of Tables.docx Hydraulic characterization and removal of metals and nutrients in an aerated horizontal subsurface flow “racetrack” wetland treating oil industry effluent List of Tables Table 1. Average influent (IN) and effluent (EFF) concentrations (μg/L), respective treatment performance (Removal, %) and inlet mass load (g/m²/yr) of the pilot RW units at the HRT of 1.37 days (n=9). Table 2. Average influent (IN) and effluent (EFF) concentrations (μg/L), respective treatment performance (Removal, %) and inlet mass load (g/m²/yr) of the pilot RW units at the HRT of 3.16 days (n=9). Table 3. First-order reaction rate coefficients (k) for the racetrack wetland design and comparison with available literature data. Table 1 RW-S RW-SA RW-SP RW-SPA Inlet mass IN EFF Rem EFF Rem EFF Rem EFF Rem load Parameter (µg/L) (µg/L) (%) (µg/L) (%) (µg/L) (%) (µg/L) (%) (g/m²/yr) Al 300 110 63.3 105 65.0 100 66.7 80 73.3 2.20 As 4.37 2.43 44.4 2.15 50.8 2.25 48.5 2.16 50.6 0.032 Ce 1 0.1 90.0 0.26 74.0 0.05 95.0 0.054 94.6 0.007 Cr 6.19 1 83.8 1.52 75.4 0.48 92.2 0.62 90.0 0.045 Cu 30.58 7.76 74.6 11.95 60.9 6.21 79.7 8.0 73.8 0.224 Fe 3580 70 98.0 160 95.5 68 98.1 142 96.0 26.3 K 6510 3440 47.2 4110 36.9 1860 71.4 2080 68.0 47.7 Li 34.62 20.32 41.3 26.61 23.1 21.31 38.4 25.2 27.2 0.254 Mg 27900 22290 20.1 27050 3.0 22170 20.5 22980 17.6 204.6 Mn 130 80 38.5 30 76.9 80 38.5 25 80.8 0.953 Mo 22.29 9.33 58.1 12.83 42.4 4.62 79.3 6.58 70.5 0.163 Ni 52.96 15.95 69.9 14.5 72.6 16.01 69.8 7.87 85.1 0.388 Sc 18.53 4.32 76.7 3.73 79.9 2.9 84.3 1.8 90.3 0.136 Se 14.27 8.97 37.1 8.89 37.7 6.5 54.4 7.1 50.2 0.105 Si 10190 8630 15.3 8040 21.1 6410 37.1 7001 31.3 74.7 Sn 1.57 1.37 12.7 1.49 5.1 0.65 58.6 0.98 37.6 0.012 Sr 2480 1390 44.0 1710 31.0 1480 40.3 2025 18.3 18.2 Ta 0.25 0.13 48.0 0.17 32.0 0.11 56.0 0.16 36.0 0.002 Th 1.05 0.09 91.4 0.1 90.5 0.06 94.3 0.07 93.3 0.008 V 101 3.36 96.7 3.64 96.4 2.63 97.4 2.52 97.5 0.741 W 2.28 0.68 70.2 1.35 40.8 0.39 82.9 0.85 62.7 0.017 Zn 12820 80 99.4 180 98.6 113 99.1 128 99.0 94.0 Pb 62 1.78 97.1 1.82 97.1 0.8 98.7 1.1 98.2 0.455 Phenols 638 3.3 99.5 2.9 99.5 1.7 99.7 0.9 99.9 4.679 NO3-N 18000 12100 32.8 13900 22.8 8300 53.9 9100 49.4 132.0 NH4-N 31000 11300 63.5 20 99.9 20 99.9 18 99.9 227.4 PO4-P 20600 20600 0.0 20600 0.0 1000 95.1 1000 95.1 151.1 Table 2. RW-S RW-SA RW-SP RW-SPA Inlet mass IN EFF Rem EFF Rem EFF Rem EFF Rem load Parameter (µg/L) (µg/L) (%) (µg/L) (%) (µg/L) (%) (µg/L) (%) (g/m²/yr) Al 300 153 49.0 120 60.0 108 64.0 91 69.7 7.38 As 4.37 3.2 26.8 2.9 33.6 3.1 29.1 0.9 79.4 0.11 Ce 1 0.12 88.0 0.13 87.0 0.05 95.0 0.05 95.0 0.02 Cr 6.19 2.3 62.8 2.5 59.6 0.48 92.2 0.62 90.0 0.15 Cu 30.58 8.8 71.2 13.4 56.2 7.01 77.1 8.8 71.2 0.75 Fe 3580 91 97.5 95 97.3 71 98.0 152 95.8 88 K 6510 2,050 68.5 4,110 36.9 1,930 70.4 760 88.3 160 Li 34.62 13.86 60.0 30.1 13.1 19.5 43.7 21.5 37.9 0.85 Mg 27900 15,390 44.8 20,015 28.3 14,510 48.0 15,310 45.1 686 Mn 130 30 76.9 10 92.3 39 70.0 10 92.3 3.20 Mo 22.29 7.37 66.9 18.4 17.5 5.93 73.4 6.61 70.3 0.55 Ni 52.96 12.18 77.0 11.63 78.0 12.52 76.4 13.71 74.1 1.30 Sc 18.53 4.29 76.8 3.89 79.0 4.2 77.3 3.5 81.1 0.46 Se 14.27 5.4 62.2 5.8 59.4 3.5 75.5 3.55 75.1 0.35 Si 10190 8,040 21.1 8,005 21.4 7,011 31.2 7,000 31.3 251 Sn 1.57 0.97 38.2 1.54 1.9 0.73 53.5 1.29 17.8 0.04 Sr 2480 890 64.1 2,130 14.1 980 60.5 1,560 37.1 61 Ta 0.25 0.2 20.0 0.24 4.0 0.06 76.0 0.08 68.0 0.01 Th 1.05 0.09 91.4 0.11 89.5 0.07 93.3 0.08 92.4 0.03 V 101 3.57 96.5 3.5 96.5 3.1 96.9 3.2 96.8 2.48 W 2.28 0.79 65.4 0.79 65.4 0.79 65.4 0.79 65.4 0.06 Zn 12820 60 99.5 130 99.0 78 99.4 91 99.3 315 Pb 62 2.6 95.8 2.62 95.8 1.3 97.9 1.36 97.8 1.52 Phenol 638 7.8 98.8 4.9 99.2 18 97.2 2.3 99.6 15.7 NO3-N 18000 16500 8.3 17300 3.9 1100 93.9 2400 86.7 442.4 NH4-N 31000 17300 44.2 20 99.9 6500 79.0 20 99.9 761.6 PO4-P 20600 20600 0.0 20600 0.0 900 95.6 890 95.7 506.1 Table 3. Parameter This Study Other Studies 219–290 m/yr (1.37 d HRT, range for all units) 42 m/yr (Manyin et al., 1997) Fe 82-107 m/yr (3.16 d HRT, range for 66 m/yr (Tarutis et al., 1999) all units) 21 m/yr (SF; Tarutis et al., 1999) 169 m/yr (1.37 d HRT with 10-60 m/yr (HSF with river gravel; Sikora et al., aeration) Mn 2000) 71-88 m/yr (1.37 d HRT without 100-600 m/yr (HSF with limestone; Sikora et al., aeration) 2000) 54-97 m/yr (1.37 d HRT, range for all units) Cu 80 m/yr (SF, estimated; Kanagy et al., 2008) 23-39 m/yr (3.16 d HRT, range for all units) 81-82 m/yr (1.37 d HRT, range for all units) Ni 15-18 m/yr (SF+HSF; Sinicrope et al., 1992) 29-42 m/yr (3.16 d HRT, range for all units) 219-284 m/yr (1.37 d HRT, range for all units) Pb 104 m/yr (SF, estimated; Kanagy et al., 2008) 95-121 m/yr (3.16 d HRT, range for all units) 25-35 m/yr (SF+HSF; Sinicrope et al., 1992) Cr 52-57 m/yr 1.37 d HRT, unplanted) 20-80 m/yr (SF; Kadlec and Srinivasan, 1995) 148-168 m/yr (1.37 d HRT, 3-64 m/yr (Kadlec and Wallace, 2009) planted) 34-46 m/yr (3.16 d HRT, unplanted) 59-66 m/yr (3.16 d HRT, planted) 360-450 m/yr (1.37 d HRT, range 28 m/yr (SF+HSF; Sinicrope et al., 1992) for all units) Zn 55-121 m/yr (SF; Gillespie et al., 2000) 118-146 m/yr (3.16 d HRT, range 21-42 m/yr (SF, estimated; Dorman et al., 2009) for all units) 8-17 m/yr (HSF, estimated; Rossmann et al., 2012) 107.2 m/yr (HSF; Stefanakis et al., 2016) 96 -165 m/yr (1.37 d HRT, planted 98.9 m/yr (HSF; Tatoulis et al., 2017) with aeration) Phenols 100 m/yr (SF, estimated; Polprasert and Dan, 1994) 152-201 m/yr (3.16 d HRT, planted 76 m/yr (SF, unplanted; Tang and Lu, 1993) with aeration) 150 m/yr (SF, planted; Tang and Lu, 1993) 100 m/yr (SF, estimated; Dong and Lin, 1994) 960,6 m/yr (VF, planted with aeration; Stefanakis et 233-237 m/yr (planted with al., 2019) aeration) 589 m/yr (HSF, planted with aeration; Stefanakis, 13.8 m/yr (1.37 d HRT, unplanted, 2020b) NH4-N no aeration) 2.9 m/yr (HSF, planted without aeration; Nivala et 24.4 m/yr (3.16 d HRT, unplanted, al., 2019) no aeration) 22.4 m/yr (HSF, planted without aeration; Stefanakis et al., 2016) 75 m/yr (HSF, planted with aeration; Nivala et al., 0.9-2.0 m/yr (1.37 d HRT, 2019) unplanted) 65.1 m/yr (VF, planted with aeration; Stefanakis et 51-73 m/yr (1.37 d HRT, planted) al., 2019) NO3-N 6.0-9.3 m/yr (3.16 d HRT, 96.4 m/yr (HSF, planted with aeration; Stefanakis, unplanted) 2020b) 16-19 m/yr (3.16 d HRT, planted) 3.9 m/yr (HSF, planted without aeration; Nivala et al., 2019) Figure Hydraulic characterization and removal of metals and nutrients in an aerated horizontal subsurface flow “racetrack” wetland treating oil industry effluent List of Figures Figure 1. (a) Racetrack wetland configuration with aeration (base unit without substrate or vegetation), and (b) the unplanted and planted sections, the coarse gravel at the inlet and outlet sections, and the influent and effluent sampling points. Figure 2. Schematic representation (top view) of the planted and non-planted sections of the racetrack wetland unit and the meandric flow path. Figure 3. Tracer response curve for the targeted 1-day and 4-days HRT. Figure 4. Removal rates of heavy metals in the four pilot CW units at the two HRT tested; low (1.37 days) and high (3.16 days). Figure 5. Removal rates of phenols, ammonia nitrogen, nitrate and phosphorus in the four pilot CW units at the two HRT tested; low (1.37 days) and high (3.16 days).