The groundwater modelling undertaken for this area is a small part of the larger programme of “restoration” of the freshwater lenses. The ultimate aim of the numerical groundwater modelling is to assess potential remediation scenarios to allow an informed decision to be made by respective authorities in Kuwait on the preferred scenario. Further to the preliminary assessment of the scenarios, the model may be used for optimisation using the results of future pilot field trials. For the purpose of remediation, detection of petroleum-related hydrocarbons in fresh groundwater samples and total dissolved solids measurements above the drinking water standard (1500 mg/L) is an indication of required remedial action. The assessment of each scenario of treatment is based on estimate of time required for remediation; and the extent of removal of contaminants (petroleum hydrocarbon and salts). The model was run from March 1991 till the end of 2063. This is to allow about 50 years of predictive remediation simulation time from the end of 2013 for the remediation scenarios. The hydrocarbon contamination is assumed to have first entered the system at the time of the oil well fire fighting in 1991.
Do nothing scenario—natural attenuation
Following the calibration of the contaminant concentration at five target bores, simulation was carried out using MODFLOW-SURFACT for 50 years (beginning year 2013) to determine the likely levels of contamination in the aquifer at different locations, including the observation bores. Chemical reaction (biodegradation) has not been considered in this simulation. Not considering biodegradation will likely overestimate the extent of contamination, therefore the rate of natural attenuation will be underestimated (conservative). Large-scale withdrawal of water from freshwater lenses will also affect the prediction results through its effects on the hydrodynamic gradient, and the transport direction and rate of movement of the pollutants.
Natural, non-enhanced, mostly microbial degradation of organic constituents by which complex organic compounds are broken down to simpler, usually less toxic compounds through aerobic or anaerobic processes can take place at any contaminated site. To rely on this natural process for environmental remediation, there must be assessment of current degradation rates to determine if they are sufficient to control or degrade a contaminant plume or zone without the creation of unacceptable risk to human health or the environment (Guvanasen et al. 2009). In the absence of site-specific data, human-assisted treatment of the contaminated groundwater may be required. In the absence of high resolution investigations, the extent of contaminant migration can only be estimated.
The simulated concentration profiles for total petroleum hydrocarbon and total dissolved solids from 2013 to 2063 at the monitoring bores designated by P-18 and P-19 are presented on Fig. 4. The total petroleum hydrocarbon concentrations slowly increased from about 0.05 mg/L after approximately a five-year to peak at 1 mg/L at about 50 years, in monitoring bore P-19 (Fig. 4). Based on the time series simulation, the total petroleum hydrocarbon concentrations increased and the plumes continued to expand. There was no apparent steady state achieved. The maximum total petroleum hydrocarbon concentration after 50 years of simulation is approximately 11.5 mg/L (at monitoring bore P-18).
Scenario 1 (do nothing) suggests that the contamination plume would continually expand over time. This type of simulation helps to determine whether concentrations will migrate into the freshwater lenses (the receptor represented by the monitoring bores) above some regulatory limit. In this case, the receptors of concern are the two freshwater lenses at Al-Raudhatain and Umm Al-Aish shown as polygons (poly-lines) in Fig. 4.
Figure 4 shows the spatial distribution of total petroleum hydrocarbon in 50 years (2063). This suggests that almost half of the main Umm Al-Aish depression will be contaminated within the next 50 years and the extent of the polluted area will increase with time. In the Al-Raudhatain basin, the contamination appears to be mainly located south of the freshwater lens.
Complete source removal option
Complete removal of the pollution sources is considered not feasible due to migration of the hydrocarbons into the vadose zone to depth of at least 25 m. However, considering this option provides an assessment of its potential effectiveness and a comparison with the do nothing scenario. It may be considered the opposite end of the remediation spectrum.
For assessment purposes, the vadose zone contamination was assumed to act as the only source of contaminants. However, since lateral and vertical contaminant migration in the vadose zone and subsequent remobilisation by recharge events are considered significant, this assumption is an over simplification of the actual case.
The assumption is that removal of the hydrocarbon source at the surface will negate future downward subsurface pollution. The results indicate that within the 50 year time frame for remediation (2013–2063), the simulated reduction in concentration is not sufficient to reach the clean-up target of 0.01 mg/L of total petroleum hydrocarbon. Results further suggest that the clean-up target may not be reached within 73 years. Beyond 23 years (the year that surface input concentrations were set to zero) the concentrations decreased to approximately 1.6 mg/L after 73 years which is still above the acceptable limit (Fig. 5).
Pump and treat and reinject (PTR) typically involves the extraction of contaminant affected groundwater through a range of extraction bores, as well as treatment of the extracted water to reduce contaminant levels. The extraction of groundwater typically induces increased flow rates of groundwater and associated contaminants through the formation, accelerating the transportation of contaminants. Contaminants in the affected saturated and wetted areas of the formation are progressively washed towards the extraction well, over time reducing the total mass of contaminants in the formation as well as reducing concentrations.
The extracted water may be treated and used for beneficial uses such as irrigation, reinjection, or other suitable uses. This section focuses on the pump–treat–reinject option.
In this study, PTR involves the pumping of water from the aquifers, treatment to remove hydrocarbons (and other contaminants as required) and desalination to reduce the total dissolved solid concentration before reinjection of the cleaned water back to the aquifer. The concept of PTR involves the washing of predominantly dissolved contaminants through the affected formation by injecting clean water and extracting contaminated water and is straight forward; however, in reality making such a scheme work efficiently is complex and depends on many variables. In the first instance, this was undertaken using analytical methods (two dimensional calculations). This approach allowed the minimum requirements for a PTR system to be assessed before modelling thereby reducing the trial and error time required to realise a workable system. The preliminary outcomes of the analytical assessment were as follows:
a large number of pumping bores are required, potentially up to 51 in Al Raudhatain and 43 in Umm Al-Aish;
individual bores are required to be screened in individual aquifers;
the individual bore pumping rates (based on the specific capacity distribution) are generally low being less than 3 L/sec for Al-Raudhatain and 2 L/sec for Umm Al-Aish;
the pumping bore capture zone should not be larger than 300 m which dictates a bore spacing of not more than approximately 600 m;
treated water quality for injection should be less than or equal to 1000 mg/L TDS and 0.0001 mg/L TPH;
reinjection may require less or the same number of bores as required for pumping;
the reinjection bores should be placed within the freshwater lens; and.
there is an assumed 5% loss of water due to desalination which may need to be made up from sources outside the basins.
General set-up and parameter values
The modelling presented here was used to test the feasibility and timeframes for remediation of the contamination in groundwater at Raudhatain and Umm Al-Aish freshwater lenses using the pump treat reinject option for treated water.
Initially, it was assumed that pumping bores would be within the area(s) of contamination with the injection bores located outside to “drive” the contamination to the pumping bores. However, preliminary analytical modelling indicated issues with this design such as upwelling of saline water, plume diving and inability to control the movement of the contamination plumes.
Given the aim is to remediate the freshwater lenses which occur over two aquifers and contamination recharge is mainly via overland flow and vertical recharge during rainfall events, the pumping and reinjection bores were required to be placed within the lenses in the area of contamination. It also became apparent that three to four times the storage volume is required to be treated to achieve adequate “washing” of contaminants from the aquifer.
The preliminary information on the layout of the PTR system from the analytical analysis was entered into the model, the aim being to achieve a reasonable capture and removal of the contaminant mass. Initially, extraction bores were located throughout the freshwater aquifers to enable adequate capture of contamination depicted after 50 years. These required approximately 100 wells in Umm Al-Aish and 80 wells in Raudhatain, each pumping at approximately 140–200 m3/day. Each well then had a capture zone of maybe 300–500 m. Wells were spaced at this interval across the two freshwater fields so as to capture possible pollutants emanating from the source areas. This requires many simulation runs to achieve convergence of the model with the adopted inputs. Preliminary parameters used are indicated below. The location of pumping and reinjection bores is shown in Fig. 6.
TPH contours depicted in Fig. 7 show the concentration contours after 73 years (50 years of PTR). The preliminary result shows that the PTR scenario is a viable option. A reasonable capture and removal of the contaminant mass has occurred via this simulation compared to the do nothing scenario. The relative reduction in the area of the 1500 and 0.01 mg/L contours (for both Raudhatain and Umm Al-Aish) is approximately 35 and 30%, respectively, although not all the TDS/TPH is removed after 50 years of pump and treat.
However, since residual pollution remains in the aquifer, some greater level of optimisation is required to optimise placement of bores and adjust pumping and re-inject rates to achieve more effective capture and removal of the contamination and hence significantly improve the clean-up. Further optimisation may allow assessment of reasonable timeframes to achieve technical remediation. Based on the preliminary runs, the maximum yielding capacity of any bore is about 450 m3/day (5.2 L/sec).
Preferred remediation option
Three options have been considered in this study through simulation:
The do nothing option will not meet the remediation targets as there are likely large sources of contamination within the vadose zone which will be mobilised over time.
The source removal option is not considered viable (or beneficial) as removal of the top several metres of dry and wet oil lakes will likely not remove the larger source of contamination which is in the vadose zone and reaches depths of at least 25 m. In addition, this remediation option may likely result in a spike in contaminant concentrations post-removal, due to increased recharge and mobilisation. On the positive, removal of the lakes will speed up natural remediation of the lakes and underlying land in the long term and may be considered as part of a larger remediation strategy.
Pump–treat–reinject, based on the preliminary runs, indicates the method may be effective in remediation of the freshwater aquifers. Therefore, PTR is the preferred remediation option (based on the modelling).
For future optimisation, a physically based management optimisation (PBMO) is suggested as the preferred tool because it integrates physics-based groundwater flow and transport models, management science and nonlinear optimisation tools to provide stakeholders with practical, optimised bore placement locations and flow rates for remediating contaminated groundwater at complex sites (e.g. Deschaine et al. 2013) where the algorithm implementation, verification and effectiveness testing was conducted, in comparison with other optimisation tools which require multiple simulation stops and starts, unable to solve complex problems in reasonable time frames, and have embedded flow and transport simulators with limited capabilities, accelerates site closure, and achieves cost savings and minimises long-term liabilities (e.g. Deschaine and Lillys 2011).
To fully assess the effectiveness of the pump and treat and reinject option further optimisation of the PTR and design of the field trial system, optimisation of monitoring bores, both current and planned, PTR pilot field trials, recalibration of the model and assessment of the results to confirm effectiveness and optimisation of the expansion of the pilot trial for further trials or full-scale remediation are required.
The pump and treat and reinject scenario although appearing to be effective will require further optimisation of the model over many model runs, and with field pilot trials in order confirm the feasibility of this option. Future optimisation of the PTR option will refine the bore layout and pumping/reinjection rates to maximise the simulated clean-up, including assessing how much of the contamination has been removed from the aquifer and weather ongoing pumping will only marginally improve remediation over the extra estimated years of pumping. Regardless, field pilot trials are required to advance this option to the feasibility stage.