Identifying and Treating Groundwater Contamination Under Infrastructure
Identifying the source of contamination beneath industrial plants can be a challenging and costly endeavor. Groundwater investigation programs often require a variety of sophisticated techniques to detect and locate contamination, which can have migrated over several decades before being detected.
Once identified, remediating the pollution can take a significant amount of time, and the effectiveness of the remediation process is highly dependent on a conceptual model of the ground conditions, including groundwater flow and solute transport.
In this article, we will explore the two conceptual models that apply to mills and refineries: the coastal model and the hill model. We will discuss the ground conditions and their implications for identifying and remediating contamination in these environments, as well as the remediation technologies available.
When creating a strategy for groundwater contamination, the type of contamination and the resultant chemistry affect the remediation strategy chosen.
Groundwater contamination may be of three different types:
- Organics (includes miscible)
- Tailings dams
- Pickled liquor solutions
- Stockpiles of product and waste
- Light non-aqueous phase liquid: also known as free product or phase separated hydrocarbons (PSH)
- Essentially is limited to hydrocarbons and can be stratified, fuels, oils solvents etc.
- Dense non-aqueous phase liquid
- Organic such as tar, bitumen
- Coke ovens,
- Gas works
- Tar or Bitumen storage tanks
Development of conceptual ground models for mills and refineries
Identification of the source of pollution beneath operational plants by groundwater investigation programs can often be as expensive and as difficult as remediation as a variety of sophisticated geochemical and geophysical techniques are often required. The long lead time between pollution event and migration to the detection point often makes identifying the source difficult.
For example, after identifying a plume under a smelter sourced from the upgradient fertiliser factory, the fertiliser management stated that they had not had a spill for at least several months. Groundwater velocity indicated that the plume we had detected commenced at least 30 years earlier. As these problems take a long time to develop, they also can take a long time to remediate. The time it takes to remediate and the effectiveness of remediation depends on the conceptual developed for the site.
Prior to undertaking remediation a conceptual model of the ground conditions including the groundwater has to be prepared. Preferred pathways caused by back fill in service trenches, and barriers to floor such as basements and deep foundations need to be incorporated into the model. Remediation planning cannot commence until a specific site conceptual model has been derived. This may also include running a groundwater flow model and solute transport model to evaluate different treatment options.
When infrastructure remains on site an understanding of the effect of barriers and preferential pathways is essential.
Coastal vs. Inland Mills
Generally mills and refineries are located according to two conceptual models. The first type of conceptual model is applicable to coastal environments. In such environments the infrastructure is underlain by permeable fill or natural sand overlying alluvial mud or coffee rock horizons which act as aquitards. Beneath the aquitard layers a semi confined aquifer frequently exists.
The second conceptual model is relevant to inland mills or refineries created by a cut and fill operation into an existing hill. Thus part of the facility is located on bedding sand over rock with the remainder on cut material located over natural soil, which in turn overlies rock. With the exception of a few modern refineries constructed in the past decade, most refineries and mills have occupied the same site since prior to 1980
Both models have similar ground conditions for the upper unconfined aquifer, but different controls on gradient. An unconfined aquifer, often as a perched water table, forms in the fill above the confining layer under the majority of refineries and mills. Contamination within this fill migrates to the perched water table and then usually migrates with the groundwater laterally to the edge of the fill, which is often a waterway and occasionally migrates vertically to the underlying aquifer, which often has water supply capabilities.
For the coastal model the gradient is slight but the permeability is high and for the hill model the gradient is normally great but the permeability low. Thus for both models, groundwater velocity is such that it can take many years to travel a hundred metres and thus a contamination plume can be insidiously spreading undetected for years.
The underlying aquifers differ markedly for the two models. The coastal model usually has an aquitard separating the upper unconfined aquifer from the lower semiconfined aquifers and contamination rarely reaches these aquifers.
In contrast, the hill model has the fill and the fractured rock aquifer usually well connected and contamination, due to the tortuosity of the rock, can often penetrate deep into the aquifer.
Usually prior to groundwater remediation beginning, the primary source is located and removed. The secondary source being the contaminated soil associated with the primary source should also be removed.
Groundwater remediation undertaken with infrastructure in place renders many treatments less effective than if structures were absent. However, the advantages of commencing remediation whilst the plant is operational include the reduction in the long term cost of remediation, and a shorter lag time after decommissioning.
Groundwater remediation technologies involve the following:
- Natural attenuation
- Enhanced natural attenuation
- Pump and treat
- Permeable reactive barriers
- Insitu treatment
- Chemical stabilisation
- Vapour extraction
- Biological degradation
Natural attenuation of inorganic contaminants
Numerous examples of unmonitored natural attenuation have occurred in the Australian mining industry. The tailings dams of Mary Kathleen Uranium are an excellent example of unmonitored natural attenuation. The tailings dams and ponds were built across an alluvial creek. The alluvial gravels at the base of the creek were highly permeable. Emplacement commenced in 1956 and was completed in 1983, with a 10 year period being under care and maintenance. Hot acidic water with a pH of below 1.0 and TDS of over 80 000 mg/kg was discharged with the tailings. The water contained per cent levels of many metals and was enriched in radium and thorium. Discharge springs had a TDS of approximately 8 000 mg/L and a pH of approximately 7 and contained very high levels of bicarbonate compared to the tailings water. Upon investigation an acid front had found to have only migrated 30 m in groundwater. Based on groundwater velocity the plume should have migrated several kilometres in the basal gravels down stream.
The soil of the surrounding area was a calcareous red earth, with a high CEC and rich in calcrete. Hydrogen sulfide was being vented from the base of the tailings. A large natural bioreactor had developed that worked in conjunction with base neutralisation and ionic exchange to attenuate the leachate plume. In front of the leachate plume was a plume of displaced native cations, and soil/leachate reaction byproducts.
Though constructed with more modern techniques and planning, seepage from Nabarlek Uranium Mine was also found to be attenuated and to have a front of attenuation products preceding the contamination plume.
Most inorganic constituents are attenuated and their attenuation rate (known as the retardation factor) can be measured in the planning stage or used to assess rate of migration of existing plumes. All chemically attenuated plumes are preceded by attenuation products. These products together with reduction in concentration both laterally and traversely are the proof that for inorganic contaminants natural attenuation alone may be sufficient.
Natural attenuation of organic compounds
Similarly to inorganic attenuation, organic attenuation is based on chemical attenuation, physical dilution, dispersion and biological degradation. However for nonchlorinated organic compounds the key mechanism is degradation. The degradation of most no-polar organic compounds in aquifers is limited by the solubility of these compounds in the water. This often leads to storage of organic contaminants on solid surfaces within the soil and results in slow release over time. Thus the key component for organic contaminants is to prove degradation. This is evidenced by reduction in concentration of the contaminants, a reduction in electron acceptors, a reduction in nutrients, a reduction in redox potential (pe) and an increase in dissolved carbon dioxide and biological activity.
Regulatory authorities require proof of degradation and ongoing monitoring programs prior to accepting natural attenuation as a remediation strategy. Usually it is limited to petrol and solvent plumes, though the strategy has been accepted for diesel at a refinery in NZ.
Enhanced natural attenuation
When natural attenuation is occurring but at a rate that is too slow to prevent the plume spreading or reaching an offsite receptor, the process can be enhanced by the addition of additives.
For hydrocarbons, this usually involves providing electron acceptors, which depending on the environment could include peroxide, air, sulfate, phosphate, ferrous iron, etc. For natural attenuation using reduction such as sulfate, chromium, heavy metals as (sulfides) add soluble organic carbon, iron chloride, sodium pyrosulfite depending on the circumstances.
Pump and treat
Pump and treat was the preferred method for groundwater treatment until the last 5 years when it was realised that such a strategy could never achieve complete remediation as a smear zone remained. It was also realised that the cone of depression caused by the pumping extending the smear zone deeper into the aquifer. However pump and treat or suck and treat aimed only at LNAPL or DNAPL compounds combined with other remediation methods is the best means of treating these separate phase layers.
When the base of the LNAPL is within 4 m (preferably 3 m) of the surface a suction extraction system can be installed targeted exclusively at the LNAPL. For deeper scenarios specialised pumps can be used that do not depress the water table. Once the free product layer is removed insitu bioremediation is usually undertaken.
Any pump and treat requires defining the groundwater parameters and modelling to define the ideal spacing of each bore based on zone of influence. Geotechnical costs for this method are expensive.
Service corridors for the operation of the system can interfere with plant operation.
Active containment is achieved by managing the groundwater gradient. Usually this means reversing the gradient at a local level. This is achieved by extraction at one point and reinjection at another. The limitations and costs for pump and treat also apply to this technique.
Passive containment involves the installation of very low permeable barrier such as bentonite slurry wall, HDPE liner or interlocked sheet piles. Tends to be effective for DNAPLs and LNAPLs but less affective on the dissolved phase, as depressurisation has to be considered. Apart from monitoring, the method has minimal maintenance requirements, however it is usually very costly to install.
Instead of active containment sometimes interception trenches are installed. Interception trenches are usually limited to unconfined aquifers overlying an aquitard, where the aquitard is within 10 metres of the surface.
The interception trench is constructed of material at least two order of magnitude more permeable than the aquifer, usually graded river gravel is used. The trench is lined with shade cloth prior to placing the gravel. The trench can be designed to collect all DNAPL, but its essential purpose is only to remove sufficient water to locally reverse the gradient so that no water leaves site.
Usually if the difference between the downgradient and the water level in the mid point of the trench is less than 100 mm (with the trench being lower) pumps at the sumps are automatically switched on.
Permeable reactive barriers
Permeable reactive barriers are permeable barriers that react with the groundwater to remove or treat the contaminants. Barriers are normally installed into excavated trenches. Chlorinated solvents are treated by this technique using a permeable granulated iron wall. Dissolved hydrocarbons can also be treated on barriers comprising peat, gravel and slow release fertilisers.
The barriers have the advantage of preventing the spread of a plume from an area that is too difficult at the time to remediate. The main disadvantage of the system is that the reactant will not last forever, clogging can occur and only a narrow range of contaminants have currently been treated by this method.
Insitu Treatment: Chemical stabilisation
Chemical stabilisation is usually undertaken in conjunction with pump and treat. Water is removed from the plume treated at the surface amended with the reagents and reinjected either as a percolation through the vadose zone and or into a series of injection wells or infiltration trenches around the edge of the plume.
This treatment can be used to immobilise chromium VI, dissolved metals by reduction, and organics including cyanide compounds by oxidation.
Oxidants include Fentons Reagent, potassium permanganate and hydrogen peroxide. Reductants include molasses, starch and sugars, iron chloride and sodium pyrosulfite.
Insitu Treatment: Vapour extraction
When used solely to remove volatile contaminants vapour extraction is limited to essentially petrol, solvents and cyanide. Modelling has to be undertaken to establish the zone of influence and thus vapour extraction is usually limited to more permeable unconfined aquifers
Vapour extraction is also used to enhance biodegradation in the capillary fringe, through removing stale air and sucking oxygen enriched air from surrounding soil into this region.
Insitu Treatment: Biological degradation
Insitu biodegradation can be achieved for nitrate, fuels, solvents, tars and other hydrocarbons and lower molecular weight PAHs. It is usually only undertaken after separate phase product is removed. Intrinsic bacteria are stimulated by the introduction of nutrients and an electron acceptor. In some instance microbes are introduced but rather than specific population being induced, diverse population such as those arising from liquid green cattle manure are most effective.
Summary of treatments
Prior to commencing remediation, the need for remediation and the clean up criteria have to be decided. Groundwater remediation is expensive and the most appropriate techniques needs to be resolved after extensive investigation and planning. Particularly as the wrong technique will often not achieved the desired results or time frame. It is worth noting that groundwater remediation time frame is measured in years, not months.
A variety of field proven techniques are available replacing or augmenting the more traditional pump and treat. Pump and treat alone has difficulty completely remediating all groundwater and usually takes years longer than planned. Groundwater remediation under infrastructure due to the complicated environment usually involves more than one of these remediation techniques. The techniques chosen depends on the chemistry of the contaminant, the physics and chemistry of the aquifer and physical constraints
There is a general perception in our community that big businesses operate outside of the oversight of the regulator, or are not scrutinised to a degree that enacts change. As Environmental, Social and Governance (ESG)...