Abstract
Abstract Two open-loop, minewater-based, ground source heating schemes have been operating since 1999–2000 in Scotland (UK), at Shettleston (Glasgow) and near Cowdenbeath (Fife). In both cases, ferruginous water is pumped from flooded coal mine workings via a borehole, circulated through the evaporator of the heat pump and reinjected via a shallower borehole to Carboniferous strata. The heat is delivered to a water-based thermal store providing central heating for apartment complexes and contributing to domestic hot water. It is demonstrated, via hydrochemical analysis and speciation modelling of the Fife minewater, that the success of the schemes is due to lack of contact between minewater and the atmosphere (thus limiting degassing of CO2 and absorption of O2). Indeed, recent difficulties with one of the schemes are ascribed to vandalism of the recharge main allowing access to oxygen, causing precipitation of ferric oxyhydroxide and clogging of the recharge borehole.
This paper is one of a thematic set entitled ‘Hydrogeology and Heat Engineering’. The use of groundwater, abstracted via open-loop systems, is increasingly being considered as a means to provide heating and cooling for buildings, particularly in London. To avoid unnecessary repetition between papers in the set, the background to the science of thermogeology and the exploitation of ground source heat has been described by Banks (2009) and the UK regulatory environment by Fry (2009).
Heat pump technology allows us to extract low-temperature ambient heat from the environment, to upgrade it to high-temperature heat and to use it for space heating. Heat pumps that extract heat energy from the geological environment (rocks, soil, sediments or groundwater) are typically referred to as ground source heat pumps (GSHP). Such GSHPs do not require pristine, uncontaminated ground conditions: heat can be extracted from contaminated soils or contaminated groundwater, providing an excellent way of ‘rebranding’ a polluted site as a source of ‘renewable energy’.
One of the most attractive potential applications for GSHP technology in Britain is the extraction of heat from minewater from abandoned (or even active) mines. The UK Coal Authority is currently responsible for managing in excess of 3000 l s−1 water from over 50 abandoned coal mine sites across England, Wales and Scotland. Assuming a coefficient of performance of 3.5 and a temperature drop at the evaporator of 5 °C, heat pump systems could deliver c. 88 MW of space heating power on the basis of this water flux: 
(1)
Similarly, in the vicinity of Katowice, Poland, the state mine dewatering organization CZOK pumps an average of 2655 l s−1 water from abandoned coal mines of the Upper Silesian Coal Basin, representing a potential space heating effect of 78 MW (Janson et al. 2009).
Such minewater discharges are usually regarded as potential sources of pollution because of the typically elevated concentrations of dissolved iron, manganese (and sometimes aluminium) and sulphate resulting from the pyrite oxidation reaction: 
(2)
Indeed, the UK Coal Authority has a rolling programme of constructing minewater treatment schemes at the most potentially polluting coal mine discharges. Although such minewater treatment schemes are often based on wetlands and are designed to be ‘passive’ as far as possible, all of them will have continuing maintenance requirements, some of them incur continuous pumping costs and a few of them require active chemical dosing (Banks & Banks 2001). They thus represent a continuing environmental liability for the Coal Authority and incur a continuing operational cost. However, many former colliery sites are now being redeveloped as commercial or residential sites (and some as museums), all with heating and cooling requirements. The possibility of using minewater as a source of heat and ‘coolth’ for such new developments offers the Coal Authority an opportunity, not only to parade its environmental credentials by converting ‘orange minewater’ to ‘green energy’, but also to recoup some of its operating expenditure by selling heating and cooling services ( Table 1). In Britain, the underground mining industry is almost defunct, but in operational mines, heat could feasibly be extracted from mine wastes or minewaters to heat mine buildings, pithead bath-houses or even to regulate the temperature and humidity of ventilation air.
Temperatures of minewater discharges from several English abandoned lead and coal/gannister mines
The idea of extracting heat from minewaters is not new. Banks et al. (2004) have demonstrated that the concept is feasible and that it has been applied in practice in the USA, Canada and Norway. Further feasibility studies and compilations of case studies have been published by Watzlaf & Ackman (2006) and Fraga Pumar (2007). In the Netherlands, a pilot project is under way to provide district heating and cooling from coal minewater in the municipality of Heerlen (http://www.minewaterproject.info/). In Poland, feasibility assessments have reached a relatively advanced stage (Solik-Heliasz & Małolepszy 2001; Małolepszy et al. 2005), in part because of the relatively high temperatures of some of the waters pumped from deep mines in the Upper Silesian Coal Basin (>20 °C from Gliwice, Szombierki, Siemianowice, Dębieńsko and Powstańców Śląskich mines; Janson et al. 2009).
In fact, the UK and Poland have ideal economic and geological conditions to develop minewater as a renewable energy source. We must therefore ask: why haven't the mining authorities or private site owners actually built such a mine-sourced heating scheme yet? One reason is possibly related to the observation that coal minewater is prone to precipitation, on the beds or rivers and in wetlands, of iron and manganese oxyhydroxides by the reactions 
(3) and 
(4) It is feared that this same hydroxide precipitation reaction may also take place within the heat exchangers of a heat pump system, causing rapid clogging and the need for an onerous programme of maintenance.
Such ‘clogging’ and a concomitant occasional maintenance requirement has been documented in the case of some open-loop GSHP schemes, based on ‘normal’ groundwater, such as that at Gardermoen International Airport in Norway (Midttømme et al. 2008). It may, therefore, come as a surprise that two operational coal minewater heat pump schemes are functioning in Scotland at present and have been doing so (relatively successfully) for 8–9 years.
The development of mine-sourced heating in Scotland
The idea of using minewater in abandoned, flooded collieries as a source of heating was considered in the early 1990s in Scotland. Fortunately, the idea was not left on the drawing board, but was put into practice in a pilot project at Mossend ([NS 74 60], Fig. 1), near Ravenscraig Steel Works, in Lanarkshire in around 1992 by an enthusiastic group of professionals including the engineers S. Johnston (of EnConsult, Dunfermline) and W. Goldie, and the geologist T. Burke. In this pilot scheme, a borehole was drilled to access flooded mine workings and the water was pumped through a heat pump at the surface, with the heat produced being used for space heating in a portable cabin (S. Johnston, pers. comm.). The scheme is no longer in use.
Outline map of Scotland, showing the localities named in the text.
Full-scale minewater-sourced heat pump schemes
The Mossend trial provided confidence that a full-scale minewater-based heat pump scheme would be viable. The group of specialists collaborated closely with John Gilbert Architects of Glasgow to commission, in 1999–2000, two new minewater heat pump schemes at housing association apartment complexes in Shettleston, Glasgow and Lumphinnans, Fife (Fig. 1). The two schemes share several features, which are described in this section. The subsequent sections detail the peculiarities of each scheme and the operational experiences.
The schemes use abstraction boreholes drilled to depthsof c. 100 m (Shettleston) and 172 m (Lumphinnans), respectively, and fitted with plain casing and c. 150 mm plastic well screens (S. Johnston, pers. comm.) in the productive horizon. Both boreholes were constructed into flooded coal mine workings from which water is abstracted by an electric submersible pump. The minewater then passes through a steel gauze filter to remove particulate matter before entering the evaporators of the heat pump array. The cooled minewater exits the evaporator(s) of the heat pump(s) and is then reinjected to the ground via a shallower borehole (reported to be typically 50–60 m deep and without well screen; S. Johnston, pers. comm.). In both cases, reinjection is designed to take place through a simple open reinjection main, below the water level in the borehole. In other words, the pump–evaporator–reinjection system is a ‘sealed’ pipe, and the minewater is not permitted to come into contact with the atmosphere.
At both sites, the heat pump array comprises two ClimateMaster WE120 water-to-water units, using R-22 (CHClF2) refrigerant, with a nominal combined output of up to 65 kW at each location, running on a negotiated cheap ‘Eco’ electricity tariff. The heat pumps produce hot water at a temperature of around 55 °C, which is stored in large insulated thermal storage tanks within the building. From here, water is distributed to provide space heating via radiators and to pre-heat domestic hot water (supplementary electric immersion heaters are used to further raise the domestic hot water temperature). At both Shettleston and Lumphinnans, a portion of the thermally ‘spent’ minewater from the heat pumps was originally diverted for use as ‘grey water’ for toilet flushing in the apartments. This was rapidly discontinued at both sites, however, because of oxidation and precipitation of iron from the minewater causing fouling and staining of the toilet systems (SUST 2006).
No routine monitoring of minewater yield or quality takes place, but one would expect that a yield of around 2–3 l s−1 would be adequate to service a peak heat load of 65 kW.
Glenalmond Street, Shettleston
The Shettleston Housing Association constructed a new 1600 m2 complex of 16 apartments in the suburb of Shettleston in eastern Glasgow in 1999 ([NS 644 637], Fig. 2) and the capital cost of the heat pump scheme was covered by a Housing Association Grant.
(a, b) Two views of the Glenalmond Street housing complex, Shettleston, Glasgow. Photograph by D. Banks.
The Shettleston area is underlain by Middle Coal Measures strata of the Lanarkshire Coalfield, which have been worked locally for coal and fireclay for at least 200 years. Coal has been taken from at least six seams, of which the Upper, the Ell and the Main are the stratigraphically highest. The 100 m deep abstraction borehole is believed to penetrate abandoned workings in the Ell Seam of Westmuir Pit from the period 1845–1862 ( Fig. 3). The pump runs 24 h per day and delivers minewater at an estimated temperature of 12 °C (John Gilbert 2000; SUST 2006). The shallower reinjection borehole is located around 20 m from the abstraction borehole. Unfortunately, no documentation is available to confirm whether the reinjection horizon is another mined seam or a sandstone horizon.
Map of the Shettleston site, showing the extent of workings in the Ell seam (shaded grey), redrawn from mine plans (especially plan 345) held by the Coal Authority.
The scheme was commissioned in 1999 and there has never been any significant failure of the scheme as a concept. The scheme delivers space heating via low-temperature radiators and is generally favoured by residents in the complex (SUST 2006). An area of 36 m2 of solar thermal roof panels supplements the temperature of the thermal store, and a back-up electrical heater has subsequently been added to the thermal store in case of problems with the heat pumps (Pither & Doyle 2005). Each apartment pays £90–100 per year for heating and hot water. This figure is designed to cover the electricity consumption of the pump and heat pumps. The housing association acknowledges that it subsidizes the maintenance bill.
As regards maintenance, the in-line filter is replaced every 3 months. The material found in the filter appears to be a mixture of particulates and ochre, the quantities being greatest after heavy rainfall: significant accumulations of ferric oxyhydroxides have not, however, been observed. A sample of the material from the filter was provided for X-ray diffraction (XRD) analysis at Newcastle University (Fraga Pumar 2007), which revealed it to be composed dominantly of quartz, with some calcite, and only minor goethite (although any amorphous iron oxyhydroxide would not have been recognized by XRD). This suggests that the majority of material trapped in the filter comprises clastic particles from the Coal Measures sandstones. The only other maintenance item on the scheme to date has been the replacement of the electric submersible pump after 8 years of service. The heat pump evaporators have reportedly never required cleaning and have never become blocked.
Ochil View, Lumphinnans
Following the success of the Shettleston scheme, a similar minewater-sourced heat pump system was constructed in 2000 in connection with the refurbishment of a 1950s apartment block, comprising 18 apartments, at Ochil View, Lumphinnans, near Cowdenbeath, Fife [NT 171 928]. The area is underlain by a part of the Fife Coalfield, known locally as the Lochgelly Coalfield, which has had a long history of coal mining. Lumphinnans XI andXII collieries worked from 1896 to 1966 (Fife Mining Heritage Society 2008), with the water discharge from Lumphinnans No. XI pit being reported as at least 75 l s−1 (Burke 1999) prior to closure. The coal-bearing strata belong to the Lower Carboniferous Limestone Coal Group and at least 11 seams have been worked in the vicinity of Ochil View, of which the stratigraphically highest are the Lochgelly Black Band Ironstone, the Seven Foot Seam, the (Kelty) Main, the Jersey or Diamond Seam and the Swallowdrum (Burke 1999).
Minewater is abstracted at a reported temperature of 14.5 °C from workings via a 172.5 m deep borehole, drilled at 8 inches and lined at 6 inches (IPR 1999; John Gilbert 2001). The workings in question are believed to be those of Lumphinnans No. 1 Colliery in the Diamond or Jersey seam, abandoned in around 1944 ( Fig. 4). The shallower reinjection borehole is located around 100 m from the abstraction well. As at Shettleston, the minewater is pumped 24 h per day and the water quality and flow are not routinely monitored. A single analysis suggests, however, a pH of around 6.2, an electrical conductivity of 2560 μS cm−1, a reducing Eh and a dissolved iron concentration of 58 mg l−1 ( Table 2).
Map of the Ochil View site, showing the extent of workings in the Jersey–Diamond seam (shaded grey), redrawn from plans held by the Coal Authority (especially plans S4115 5/7 and S406 3/11).
Minewater sample taken from discharge pipe (Fig. 5) at Ochil View on 25 May 2007
The vandalized discharge main at Lumphinnans. Thermally ‘spent’ minewater cascades into a manhole; the recharge borehole itself is underwater. Accumulation of ‘ochre’ can be seen, especially around the mouth of the discharge pipe.
The combined effect of the building refurbishment and installation of the heat pump system has been to halve the primary energy cost for heating and domestic hot water per apartment and reduce the annual CO2 emission from 5.2 to 1.9 tonnes per year. Prior to refurbishment the apartments were heated by electric storage heaters (FSHA 2003), using electricity from the national grid.
The Lumphinnans scheme has been more problematic in terms of operation than the Shettleston scheme, although no critical failures can be ascribed to the system concept as such. Indeed, the gauze filters at Lumphinnans generally do not require replacement, and no significant accumulations of ferric oxyhydroxide are observed in the heat pump evaporators or in the associated pipework (S. Johnston, pers. comm.). There have, however, been operational problems related to the heat pumps themselves and, more significantly, in around 2005, vandals damaged the recharge main at the reinjection well, necessitating the well to be redrilled. Following redrilling, the recharge water is not ‘injected’ below the water level in the borehole, but allowed to cascade down the borehole from a manhole (Fig. 5). This has led to precipitation of ‘ochre’ (confirmed by XRD analysis to contain goethite; Fraga Pumar 2007) in the recharge borehole, clogging of the borehole and periodic overflowing of the borehole at the surface. The submersible pump was found on removal in 2006–2007 to be covered in ‘ochre’. The Coal Authority considers a pump lifetime of 5–10 years in ochreous mine water as typical.
The impact of ground source heat extraction on mineral precipitation
The British reluctance to use minewater in open-loop ground source heat systems can, to some extent, be ascribed to the belief that there is a risk of iron and manganese oxyhydroxides precipitating out in heat exchange elements (or in a recharge well) necessitating regular replacement or maintenance. The potential concerns can be summarized as follows.
A temperature change will alter the solubility products and saturation indices of the relevant mineral phases.
Carbon dioxide will degas from the minewater, raising the pH and making oxyhydroxide minerals (or even carbonates, such as calcite) more likely to precipitate. This phenomenon is recognized from minewater treatment (McAllan et al. 2009).
Exposure to atmospheric oxygen will oxidize ferrous to ferric iron, in turn causing precipitation of ferric oxyhydroxide minerals.
A change in temperature will increase the rate of oxidation and precipitation of minerals.
An increase in temperature will favour enhanced biofilm formation in heat exchangers or a recharge well.
The first three of these concerns can be investigated using a hydrochemical modelling program, such as PHREEQCI (Parkhurst & Appelo 1999). A sample of water (Table 2), taken from the Lumphinnans discharge main on 25 May 2007 (Fig. 5), was analysed both in the field (for pH, redox potential, temperature, electrical conductivity and alkalinity) and at the laboratory of the Hydrogeochemical Engineering Research and Outreach (HERO) at Newcastle University. The program PHREEQC Interactive version 2.13.2 (Parkhurst & Appelo 1999) has been used to simulate the speciation of inorganic components in the minewater (Table 2), using the PHREEQC thermodynamic database. It is found that the water is in a reducing condition, with iron and manganese overwhelmingly present in the FeII and MnII oxidation states. The water has a significant excess partial CO2 pressure of 10−0.72 atm, compared with around 10−1.5 atm in many ‘normal’ groundwaters (reflecting soil CO2 concentrations) and 10−3.5 atm in the atmosphere. The high partial CO2 pressure suggests that protons generated by pyrite oxidation (equation (2)) have been neutralized by carbonates in the formation (calcite, ankerite, siderite), releasing excess inorganic carbon to the groundwater. Another potential source of excess CO2 may have been sulphate reduction in reducing niches within the mine or aquifer system. The water is undersaturated with respect to ferrihydrite (taken to be Fe(OH)3(a) in the PHREEQC database), pyrolusite (MnO2) and calcite. The water is only slightly undersaturated with respect to rhodochrosite (MnCO3), chalcedony (SiO2) and gypsum (CaSO4.2H2O), and is slightly oversaturated with respect to siderite (FeCO3).
PHREEQCI has been used to simulate the potential for mineral precipitation as a result of change in temperature, degassing of CO2 and contact with atmospheric oxygen. It should be emphasized that PHREEQCI has been applied in this context only to explore thermodynamic tendencies to precipitation or dissolution; no attempt has been made to address the likely importance of kinetic factors or microbiological catalysis of such reactions. When simulating precipitation of iron oxyhydroxides in a heat exchanger or recharge well, one must consider the mineral phase that is likely to be precipitated. As it is generally accepted (see Langmuir 1997) that a semi-amorphous hydroxide (often termed ferrihydrite Fe(OH)3.nH2O) is the normal product of rapid hydrolysis of Fe3+ and/or oxidation of Fe2+, amorphous iron hydroxide (assumed to be synonymous with ferrihydrite in PHREEQCI) was selected as the critical phase for the purpose of modelling. Figure 6 illustrates that changing the temperature of the minewater alone results in little change to the calculated saturation indices. Indeed, a temperature change of ±6 °C (comparable with the temperature change occurring in most ground source heat pump systems) does not result in either calcite or iron hydroxide becoming oversaturated.
Modelled saturation indices using PHREEQCI for the Lumphinnans minewater (Table 2), varying only temperature in 2 °C increments. The term ferrihydrite is assumed synonymous with amorphous Fe(OH)3 in the context of the PHREEQCI database.
PHREEQC has also been used to simulate the progressive degassing of CO2 in the event of exposure of minewater to the atmosphere or to significant decreases in pressure within the heat pump–recharge system. The Lumphinnans minewater has been progressively equilibrated with CO2 partial pressures of 10−1.0, 10−1.5, 10−2.5 and 10−3.5 atm, the last partial pressure being comparable with the atmospheric partial pressure of CO2. The simulated degassing of CO2 leads to a significant rise in pH as protons are consumed from solution: 
(5)
This leads, in turn, to a progressive increase in the calcite saturation index, becoming saturated at PCO2 = 10−1.5 atm and significantly oversaturated by PCO2 = 10−3.5 atm. It also leads to an increase in the saturation index of ferric hydroxide, with the solution becoming approximately saturated if an equilibrium with the atmosphere (PCO2 = 10−3.5 atm) is achieved ( Fig. 7).
Modelled saturation indices using PHREEQCI for the Lumphinnans minewater, varying only partial pressure of CO2. The initial minewater (Table 2) has a partial pressure of CO2 of 10−0.72 atm; this is allowed to degas to an atmospheric value of 10−3.5 atm.
Finally, the exposure to atmospheric oxygen was simulated by progressively adding oxygen to the solution in PHREEQCI ( Fig. 8). We can see from the equation for oxidation of ferrous iron 
(6) that each mole of ferrous iron requires 0.25 moles of oxygen for complete oxidation. The Lumphinnans minewater contains around 1 mmol l−1 of ferrous iron. In theory, therefore, 0.25 mmol l−1 oxygen (8 mg l−1) are required to completely oxidize it. We can see from Figure 8, however, that after the addition of only a few mg l−1 oxygen, sufficient ferrous iron has been oxidized to the ferric form for ferric hydroxide to become oversaturated and potentially precipitated. As more oxygen is added, the ferric hydroxide saturation index increases further. After 8 mg l−1 oxygen has been added, the iron is fully oxidized and the ferric hydroxide saturation index increases no further. At this point, oxygen is consumed in oxidizing manganese (+II) and pyrolusite (MnIVO2) rapidly becomes saturated.
Modelled saturation indices using PHREEQCI for the Lumphinnans minewater, varying only exposure to O2. The initial minewater (Table 2) is reacted with O2 in 0.05 mmol l−1 (1.6 mg l−1) increments.
As regards the kinetics of iron oxidation and precipitation, the following observations can be made. The rate of abiotic oxidation of ferrous iron by oxygen (equation (6)) is known to be first order with respect to iron and oxygen, and second order with respect to OH−, at pH > 5 (Singer & Stumm 1970; Stumm & Morgan 1996): 
(7)
At pH < 4, the abiotic rate becomes independent of pH. The abiotic oxidation rate increases about tenfold for every 15 °C temperature increase (Sung & Morgan 1980; Stumm & Morgan 1996). At low pH, the presence of iron-oxidizing microbes also becomes critical and can result in several orders of magnitude more rapid oxidation. Biotic oxidation rates decrease with increasing pH, however, and become less significant than abiotic oxidation above pH 5–6. Again, overall (biotic + abiotic) oxidation rates increase significantly with temperature (Kirby et al. 1999).
Conclusions
As a result of PHREEQCI modelling and consideration of kinetics, we can reach the following conclusions regarding the tendency of the Lumphinnans minewater to cause chemical clogging in heat exchangers or a recharge well.
Pressure reductions in the pumping system and/or exposure to atmospheric carbon dioxide concentrations have the potential to result in degassing of carbon dioxide from the minewater. This results in a potential hazard of CO2 gas accumulation in confined spaces and in a rise in pH of the minewater, leading to a thermodynamic tendency for the precipitation of calcite and iron hydroxide. Of these, ferric hydroxide precipitation would be expected to be the greatest potential hydrochemical problem, as its precipitation is surface catalysed and can be potentially rapid; moreover, it is rather voluminous in its amorphous forms.
Exposure to only a small amount of atmospheric oxygen can lead to ferric hydroxide (and even pyrolusite) becoming significantly oversaturated and potentially precipitating.
Net heating schemes, which result in chilling of the abstracted groundwater, would be expected to result in a reduction in the rate of biotic and abiotic iron oxidation (if thermodynamically favourable) and probably also in the rate of biofilm formation. Net cooling schemes, where heat is dumped to the groundwater flow, would be expected to result in an increase in the rate of iron oxidation and biofilm formation.
If, however, the pumped minewater is not exposed to atmospheric oxygen or carbon dioxide concentrations, nor to significant changes in pressure, the change in temperature of the minewater alone, as it passes through the heat exchanger, is not sufficient to result in significant changes in saturation indices of potentially precipitating mineral phases, nor in oversaturation of calcite, pyrolusite, chalcedony or ferric hydroxide.
Thus, if pumping does not result in dewatering of the flooded mine horizon and admixture of oxygen, if the pipework and heat exchanger are sealed units, if recharge takes place well below the water level in the recharge well and if pressures are regulated in the system (e.g. by regulation valves on the discharge main in the recharge well), the potential for precipitation of ferric and manganese oxyhydroxides and of calcite should be minimized.
These conclusions wholly support the operational experiences from Shettleston and Lumphinnans; that is, that significant mineral precipitation has not been observed in the pipework, heat exchanger or recharge wells under normal operation. Following the vandalism of the recharge main at Lumphinnans, however, such that admixture of oxygen has occurred, significant iron oxyhydroxide precipitation has occurred within the recharge well.
Although the findings of this study cannot automatically be extrapolated to other mine systems and water chemistries, they offer grounds for optimism that, by careful management of system pressures and minimizing exposure to atmospheric conditions, minewater-based GSHP schemes can function reliably and with minimal maintenance requirements.
Acknowledgements
The authors would like to thank the following for their assistance in facilitating visits and sampling of the sites in Fife and Glasgow: R. Bell and J. Whitley of the Shettleston Housing Association; R. Milne and N. Engels of the Fife Special Housing Association; and S. Johnston of EnConsult Ltd. in Dunfermline. This paper was also presented at the Minewater '08 conference in Aachen–Heerlen in October 2008.
- © 2009 Geological Society of London
The operational performance of Scottish minewater-based ground source heat pump systems
