Performance and Review of Passive Minewater Treatment Sites, Valley, Alyson Bannister1,2, Peter Brabham2, Tim Jones2, Rob Bowell3

1SRK Consulting (South Africa), Illovo House 2146, Johannesburg, South Africa 2Department of Earth Sciences, Cardi University CF10 3AT, Cardi , 3SRK Consulting (UK), Churchill House CF10 2HH, Cardi , United Kingdom

Abstract A series of  ve constructed passive wetlands were constructed between 1995-1999 in the Pelenna valley, South Wales to treat various minewater discharges from abandoned coal mines. Two of the wetlands, Whitworth No.1 and Garth Tonmawr, have been com- pared to assess their long-term performance over their 20-year design life as well as their individual cell performance. e wetlands have not performed exactly as designed but over the long-term they have achieved the required remediation targets of pH 6-9 and dissolved iron of less than 1 mg/L in the Afon Pelenna downstream of the wetlands. It is concluded that the more complex the cell design the less likely the cell will operate for its design life and more maintenance will be required. Keywords: oxidation | hydrolysis | aerobic | anaerobic | reducing and alkalinity produc- ing system

Introduction ing the cessation of active mining operations, Coal mining in the United Kingdom (UK) the coal workings  ooded and minewater dis- has gone through a full rise and fall cycle, out- charged into the river system, Nant Gwen rwd put rose constantly through the 18th century and Nant Blaenpelenna, tributaries of the Afon reaching a zenith output around 1913. e Pelenna. e two tributaries were stained or- 20th century saw a continual decline until the ange with iron concentrations elevated for ap- last deep coal mine closed in 2015. In 1994, it proximately 7 km downstream to the con u- was estimated that 200 km of UK rivers and ence with the (Edwards et al. 1997). streams were a ected by coal minewater dis- e aquatic species were thought to be impov- charge (NRA 1994). erished from a combination of increased acid- During active coal mining, the under- ity, toxic e ects of metals and from the smoth- ground workings are “dewatered” to maintain ering e ect of ochre on the benthic zone of the a safe working environment, and to reduce watercourse (Wiseman 2002). water ingress creating mostly dry conditions In 1992, a study was initiated to establish which prevent the mobilisation of contami- the impact that minewater discharge was nants. Once pumping ceases and the water having on the environment. e recommen- table rebounds, pyrite is exposed to oxygen, dations from the study were to reduce iron water and bacterial catalysts, and the oxida- concentrations by 95% and 50% in the Nant tion of pyrite is stimulated producing acid Gwen rwd and Nant Blaenpelenna respec- mine drainage. Ideal conditions are created in tively, so that iron concentrations would be abandoned workings for the oxidation prod- below the required 1 mg/L and the pH would 2+ 2- ucts, ferrous iron (Fe ), sulfate (SO4 ) and be between 6-9 in the Afon Pelenna. is free acidity, to be dissolved and carried out was expected to provide suitable conditions of the mine as minewater discharge into the for recolonisation by salmonid  sh (Ishemo surrounding watercourses. and Whitehead 1992). It was decided that the Minewater discharge from pyrite oxidation most suitable and cost-e ective remediation may have serious environmental consequences of the tributaries would be to passively treat for aquatic ecosystems as it did in the Pelenna the minewater discharge through construct- valley in the South Wales Coalfi elds. Follow- ed wetlands.

84 Wolkersdorfer, Ch.; Sartz, L.; Weber, A.; Burgess, J.; Tremblay, G. (Editors)

02_Water treatment process BOOK.indb 84 9/3/18 11:55 AM 11th ICARD | IMWA | MWD Conference – “Risk to Opportunity”

Between 1995 and 1999, a three-phase e three-phase passive wetland scheme passive wetland scheme was constructed to was constructed to treat minewater discharge treat  ve minewater discharges; this was one in the Pelenna valley. is paper only dis- of the  rst passive minewater treatment wet- cusses Whitworth No.1 and Garth Tonmawr lands within the UK and Europe. Diff erent as they have comparable constructed wetland passive constructed wetland con gurations designs and are of similar age. were implemented based on the incoming minewater discharge. Whitworth No.1 e con gurations used were considered Whitworth No.1 was the  rst phase, com- novel as they were mainly based on work at pleted in October 1995 and comprised of four the time in the USA of Hedin et al. (1994) and parallel cells. e incoming discharge was Kepler and McCleary (1994). Th e scheme was split into each cell of which two were aero- known as the River Pelenna Minewater Proj- bic and two were anaerobic. e aerobic cells ect before the Coal Authority took over man- were expected to remove iron by oxidation agement of the scheme that is now known as and hydrolysis while iron was expected to be the Coal Authority Pelenna Minewater Treat- reduced by sulfur reducing bacteria (SRB) ment Sites (MWTS). in the anaerobic cells. Di erent substrates and vegetation types were utilised for each Site Description cell; this was largely due for experimental e Pelenna MWTS consist of  ve construct- purposes with the results used to inform the ed wetlands, Whitworth No.1, Garth Ton- construction of phase II and III. e wetland mawr and Whitworth A, B and Gwen rwd, is constructed in precast concrete with a geo- located near Tonmawr village, South Wales synthetic basal liner (SRK 1994). approximately 11 km northeast of Port Tal- e design of Whitworth No.1 changed be- bot. Tonmawr lies within the Pelenna valley tween 2006 and 2009 to allow the cells to  ow through which the Nant Gwen rwd and Nant in series from cell 4 to cell 1, due to a reduction Blaenpelenna  ow into the Afon Pelenna. e in wetland performance. Cell 1 and 4 became Afon Pelenna is a tributary of the river Afan, settlement ponds while cell 2 and 3 became which  ows into the sea at . aerobic wetlands as illustrated in Figure 2.

Figure 1 Schematic of Whitworth No.1 in 2017

Wolkersdorfer, Ch.; Sartz, L.; Weber, A.; Burgess, J.; Tremblay, G. (Editors) 85

02_Water treatment process BOOK.indb 85 9/3/18 11:55 AM 11th ICARD | IMWA | MWD Conference – “Risk to Opportunity”

Figure 2 Schematic of Garth Tonmawr in 2017

e Garth Tonmawr constructed wetland PHREEQC was used to determine the re- is illustrated in Figure 3 and consists of a set- dox potential (Eh) of the minewater at the in- tlement lagoon, two reducing and alkalinity lets and outlets. e initial run produced a list producing systems (RAPS) cells and two an- of possible species present within the water aerobic cells. e settlement lagoon promotes and an estimated concentration of each spe- iron oxidation and hydrolysis and allows cies that included Fe3+ and Fe2+, which were heavier particles to settle out before being subsequently used in the redox calculations further treated. e RAPS cells are anaero- to determine the Eh (Appelo and Postma bic and make use of a limestone bed overlain 2006). with substrate. e water moves through the substrate, which allows some precipitation of Ochre sampling and analysis metals before the limestone bu ers the acid- ree ochre samples were collected from the ity of the minewater. four cells at Whitworth No.1. e samples were prepared by drying and crushing before Methods elemental composition was analysed using an Water quality data analysis Olympus Innov-X X-ray Fluorescence (XRF) Natural Resources Wales (NRW) provided analyser. long-term water sampling data for the Pel- e samples were also acid leached and enna MWTS. is included data from river the leachate analysed using an Inductively samples up and downstream, and the inlet Coupled Plasma Mass Spectrometry (ICP- and outlet of the wetlands. MS) to determine the mobile elements re- Using the statistical programme R, the leased from the ochre under acidic condi- data was standardised by making box and tions. whisker plots and removing any outliers in Results and Discussion the data before processing with Excel. e data was reduced to 5 years prior to construc- Performance Assessment tion of each wetland until January 2017. Ac- When the Whitworth No.1 cells operated in cess was used to align the months of the long- parallel, there was an apparent increase in term data at each monitoring station to plot acidity and a decrease in the sulfate removal long-term trends and determine the removal e ciency, which may be explained by a re- rate in and out of each wetland. duction in SRB due to unfavourable condi-

86 Wolkersdorfer, Ch.; Sartz, L.; Weber, A.; Burgess, J.; Tremblay, G. (Editors)

02_Water treatment process BOOK.indb 86 9/3/18 11:56 AM 11th ICARD | IMWA | MWD Conference – “Risk to Opportunity”

tions; SRB occur under anaerobic conditions iron particulates may be released from cell 2 at medium to high pH (Bass Becking et al. and are not fully settling out in cell 1 before 1960). Whitworth No.1 made use of a  ow the discharge is released into the Nant Blae- distribution chamber to distribute the incom- npelenna. ese reactions also explain the ing minewater to each of the four cells; this increase in sulfate in the wetland as sulfate is likely increased the acidity at the inlet. It is released during pyrite oxidation and during thought that the decrease in pH at the inlet release of previously bound metal sul des. and the decreased performance of the an- It should be noted that aluminium and aerobic cells was the reason for the change in manganese concentrations are low but ap- design to series cells. pear to be su ciently removed from the wet- e dissolved iron removal rate has re- land. e removal rate of both constituents mained high and variable throughout the increased with the change in design, which monitoring period (Table 1) although the to- may be due to an increased retention time. tal iron removal rate has decreased to 76% in Aluminium precipitation occurs around pH the current period. e iron Eh-pH diagram 5 and under the net acidic conditions alumin- created for Whitworth No.1 indicates that ium oxidation is expected (Hedin et al. 1994). the conditions are suitable for iron oxidation is is supported by the Eh-pH diagram

(Bass Becking et al. 1960), with Fe(OH)3 de- where aluminium is in the oxidised form position. e iron is susceptible to spikes in Al(OH)3. According to the Eh-pH diagram, acidity within the wetland and according to manganese is in Mn2+ form suggesting that the Eh-pH diagram this would create condi- the manganese is being precipitated as metal 2+ tions for the Fe(OH)3 to be reduced to Fe . sul des rather than oxidised. Oxidation of From the 2010-2014 period, the wetland ap- manganese occurs at around pH 8 (Stumm pears to be a source of sulfate rather than a and Morgan 1981). sink. e discharge from the outlet at Whit- is decrease in performance at Whit- worth No.1 has been within the pH 6-9 range worth No.1 may be from a combination of required for the Afon Pelenna. e dissolved factors. Firstly, reduced hydraulic retention iron concentrations represented by the  l- due to the system short-circuiting through tered iron was supposed to be reduced by 95% preferential  ow paths potentially created in the Nant Gwen rwd to be below 1 mg/L during blockages or periods of high rainfall. in the Afon Pelenna. is means that the  l- Secondly, the lack of speci c oxidation tered iron concentrations at the outlet should and reduction conditions needed to remove be approximately 1.98 mg/L; this has been iron. Th e change in Cell 2 from anaerobic to achieved in the last two years but not over the aerobic may have caused the pyrite formed 22-year life, with an average of 2.34 mg/L. e during bacterial sulfate reduction, to be oxi- dissolved iron would therefore be expected to dised. is requires ferric iron to be available, be around 1 mg/L in the Afon Pelenna due to which may be the case as not all the ferrous dilution in the Nant Blaenpelenna. iron will be converted through cells 4 and 3. At Garth Tonmawr, iron oxidation and is is supported by the XRF results, which hydrolysis are occurring as designed (Table show that cell 2 is only retaining 28% iron 2), as evidenced by surface ochre on the  rst within the ochre while cells 4 and 3 are re- three cells. e dissolved iron removal rates taining above 50% and cell 4 is retaining show a small decrease over time to about 90% above 40%. Th e ICP-MS results are an indica- in the current period with oxidising condi- tion of what would be released from the ochre tions. under acidic conditions. is also supports Garth Tonmawr has a net-acidic mine- the above suggestion as 7.1 mg/g was released water discharge and so it was decided to from the sediment in cell 2 while 6.9, 6.6 and implement RAPS cells within the wetland. 6.5 mg/g were released from cells 3, 4 and 1, e RAPS cells seem to have decreased in respectively. e ciency quite quickly as the acidity in the e di erence in  ltered and total iron discharge is progressively less bu ered in the removal also supports this suggestion as the wetland.

Wolkersdorfer, Ch.; Sartz, L.; Weber, A.; Burgess, J.; Tremblay, G. (Editors) 87

02_Water treatment process BOOK.indb 87 9/3/18 11:56 AM 11th ICARD | IMWA | MWD Conference – “Risk to Opportunity”

Table 1. Summary of the long-term data at Whitworth No.1

Date Whitworth N0.1 Flow pH Total Fe Filt Fe Filt Mn Filt Al SO4 (m /s) (mg/L) (mg/L) (mg/L) (mg/L) (mg/L) 1995-1999 Inlet - 6.3 22.97 20.99 1.97 0.065 346 Outlet 0.003 7.0 3.95 3.23 1.08 0.010 298 Removal Rate (%) - - 83 85 45 85 14 2000-2004 Inlet - 6.0 24.27 22.13 1.88 0.091 334 Outlet 0.004 6.3 2.38 2.21 0.86 0.015 313 Removal Rate (%) - - 90 90 54 83 6 2005-2009 Inlet - 5.8 25.63 - - - 332 Outlet - 6.0 0.67 - - - 274 Removal Rate (%) - - 97 - - - 17 2010-2014 Inlet 0.004 6.2 19.40 17.51 1.65 0.098 271 Outlet 0.005 7.0 2.40 2.08 0.63 0.011 273 Removal Rate (%) - - 88 88 62 88 -0.6 2015-Present Inlet - 6.2 19.32 18.55 1.72 0.116 279 Outlet - 6.8 4.60 1.84 0.68 0.010 281 Removal Rate (%) - - 76 90 61 91 -0.9 1995-Present Inlet 0.004 6.1 22.32 19.79 1.80 0.092 312 Outlet 0.004 6.6 2.80 2.34 0.81 0.012 288 Removal Rate (%) - - 86.8 88.3 55.5 86.8 7.1

e decreased performance in the RAPS iron concentration is required to be reduced cells may be due to changed water levels af- by 50% in the Nant Blaenpelenna to reach the fecting  ow through the substrate and lime- 1 mg/L standard in the Afon Pelenna; this stone beds. e over ow means that some would require a dissolved iron concentration water is bypassing the cell and will not be of 14.5 mg/L, which has been consistently bu ered by alkalinity. reached over each 5-year period. e performance of the wetland has still been maintained, which may be due to the in- E ect of wetland design on performance crease in pH at the Garth Tonmawr discharge Both wetlands had high iron removal rates to above pH 6, which is suitable for microbio- and a similar capacity to bu er acidity indi- logical catalysts to increase iron oxidisation cating that both parallel and series cells can and bu er the pH (Bass Becking et al. 1960). perform well if designed correctly and main- e sulfate removal rates in Garth Ton- tained. Whitworth No.1 showed an initial in- mawr have been poor because conditions crease in performance when the wetland de- to promote sulfate reduction were not im- sign changed to operate in series, this may be plemented during the construction of the due to an increased practical hydraulic reten- wetland. Aluminium concentrations of the tion time, which allows for further chemical discharge are low, but the removal rate has and biological processes to take place. remained high (85%) even though it has de- Wiseman (2002) showed that the aerobic creased over time, this may be due to a re- cells were performing better than the anaero- duced hydraulic retention time. e manga- bic cells at Whitworth No.1. Anaerobic cells nese removal rate is low due to poor oxidising are more sensitive to  uctuations in water conditions in the wetland. Metal sul des are level from blockages than aerobic cells be- unlikely to precipitate out of the water as the cause the water will  ow under higher pres- conditions do not favour SRB and therefore sure and may move to the surface of the cell the manganese would not be expected to co- or create openings in the substrate which are precipitate to manganese sul de. likely to become preferential  ow paths or al- e discharge from the outlet at Garth ternatively, the cell may completely dry up af- Tonmawr has been within the pH 6-9 range fecting the microbes in the cell. e limiting required in the Afon Pelenna. e dissolved factor for anaerobic cells will be carbon if the

88 Wolkersdorfer, Ch.; Sartz, L.; Weber, A.; Burgess, J.; Tremblay, G. (Editors)

02_Water treatment process BOOK.indb 88 9/3/18 11:56 AM 11th ICARD | IMWA | MWD Conference – “Risk to Opportunity”

Table 2. Summary of the long-term data at Garth Tonmawr

Date Garth Tonmawr Flow (m /s) pH Filt Fe Filt Mn Filt Al SO4 (mg/L) (mg/L) (mg/L) (mg/L) 1999 - 2003 Inlet - 5.9 32.45 0.66 0.123 281 Outlet 0.019 6.7 1.20 0.48 0.010 274 Removal Rate (%) - - 96 27 92 2.3 2004 - 2008 Inlet - 5.6 - - - 290 Outlet - 6.1 - - - 287 Removal Rate (%) - - - - - 0.9 2009 - 2013 Inlet 0.016 6.2 26.43 0.58 0.088 217 Outlet 0.019 6.5 2.11 0.52 0.010 214 Removal Rate (%) - - 92 10 88 1.3 2014 - Present Inlet 0.024 6.2 28.11 0.55 0.068 213 Outlet 0.025 6.6 2.43 0.44 0.010 207 Removal Rate (%) - - 91 20 85 2.9 1999 - Present Inlet 0.020 6.0 29.00 0.60 0.093 250 Outlet 0.021 6.5 1.91 0.48 0.010 246 Removal Rate (%) - - 93 19 88.3 1.9

anaerobic and redox conditions are preserved Acknowledgements to support SRB while the limiting factor for the aerobic cell appears to be space rather I thank the Commonwealth Scholarship than the cell becoming exhausted. Commission, which funded my Cardiff Uni- e change in design at Whitworth No.1 versity Master’s thesis on which the paper is has altered the anaerobic cell to an aerobic based. In addition, I thank the Coal Authority cell. is has changed the cell from reducing and Severn Trent Services for allowing me to conditions to oxidising conditions and conse- access and collect samples from the Coal Au- quently there is evidence for pyrite oxidation thority Pelenna Minewater Treatment Sites and release of iron sul des once sorbed in the and the Natural Resources Wales for provid- cell. Cell 4, which was an anaerobic bed was ing the long-term water database. A special cleared of substrate and is now a settlement thank you to Peter Brabham, Tim Jones and lagoon which appears to be promoting iron Rob Bowell for their supervision of my thesis oxidation and hydrolysis as ochre is  oating and for their contribution to the paper. on the surface. References Conclusions Appelo CAJ, Postma D (2006) Geochemistry, Even though these wetlands are not perform- groundwater and pollution, 2nd Ed. A.A Balke- ing as designed they are still removing the ma Publishers, Leiden, Netherlands, p 415—488 dissolved iron and bu ering the pH to suit- Baas Becking LGM, Kaplan IR, Moore D (1960) able concentrations. is indicates that these Limits of the natural environment in terms of wetlands are resilient to change and that they pH and oxidation-reduction potentials. e may be somewhat over designed. It appears Journal of Geology 68:243—285 that more simplistic designs would have pro- Edwards PJ, Bolton CP, Ranson CM, Smith AC duced similar results and that the more com- (1997) e River Pelenna minewater treatment plex the cell con guration the less likely they project. In: Younger PL ed. Minewater treat- are to last the design life. is is mostly due ment using wetlands, Proceedings of a national to higher maintenance requirement for more conference. London Chartered Institute of Wa- complex cells. It is therefore concluded that ter and Environmental Management, Newcas- regular maintenance is essential for e cient tle, p 17—33 performance.

Wolkersdorfer, Ch.; Sartz, L.; Weber, A.; Burgess, J.; Tremblay, G. (Editors) 89

02_Water treatment process BOOK.indb 89 9/3/18 11:56 AM 11th ICARD | IMWA | MWD Conference – “Risk to Opportunity”

Hedin RS, Nairn RW, Kleinmann RLP (1994) e National Rivers Authority (NRA) (1994) Aban- passive treatment of coal mine drainage. US Bu- doned mines and the water environment. reau of Mines Information Circular 9389 HMSO Water Quality Series No 14 Ishemo CAL, Whitehead PG (1992) Acid mine SRK (1994) Pelenna Minewater Treatment Project: drainage in the River Pelenna: modelling and Design Review of Proposed Wetland Whitworth pollution control. Wallingford: Institute of Hy- No. 1 Discharge for West County drology Council. Report No. U497/1, SRK (UK) LTD, Kepler DA, McCleary EC (1994) Successive Alka- Cardiff linity Producing Systems (SAPS) for the Treat- Stumm W, Morgan J (1981) Aquatic Chemistry, ment of Acid Mine Drainage. Proceedings of 2nd Ed. John Wiley & Sons, New York the International Land Reclamation and Mine Wiseman I (2002) Constructed Wetlands for Drainage Conference and the 3rd International Minewater Treatment. R&D Technical Report Conference on the Abatement of Acidic Drain- P2-181/TR, Environmental Agency, Bristol age, Pittsburgh, p 195—204

90 Wolkersdorfer, Ch.; Sartz, L.; Weber, A.; Burgess, J.; Tremblay, G. (Editors)

02_Water treatment process BOOK.indb 90 9/3/18 11:56 AM