10.2 MAR site selection process
10.2.1 reducing the potential MAR sites from eleven to six
The eleven potential sites for MAR were selected to provide a reasonable spatial coverage of the study area with which to test the impacts of MAR on groundwater levels and provide an initial indication of TWW plume extent/travel direction. No consideration was given specifically to land use, environmental assets or other factors that may impact on site selection at this stage. Six sites were within Tamala Limestone and Safety Bay Sand aquifers close to the SDOOL and five were 4 to 6 km inland in less transmissive sandy aquifers.
Groundwater flow model results were presented for each of these sites operating individually at either 4.8 or 9.6 ML per day. In addition, results from combinations of sites were presented, namely pairing of sites 1 and 7; sites 3 and 9; and sites 5 and 11; all 6 coastal sites along the SDOOL, all 5 inland sites, and all of the sites operating simultaneously.
Table 10.1 shows the wastewater volumes that were applied in these model runs. To provide a frame of reference, the simulated MAR volumes were compared with the volume of wastewater discharged via the SDOOL as this would potentially be the source water for MAR. For example, operation of all 11 sites at 9.6 ML/d (2 m/d) of infiltration would recharge the aquifer with 76% of the wastewater currently discharged via the SDOOL and nearly half of the projected wastewater discharge for 2032. In contrast, the operation of only one MAR site would use only 3.4% of the wastewater currently discharged via the SDOOL and 2.1% of the projected wastewater discharge for 2032.
Table 10.1 Volumes of wastewater modelled in simulations of MAR in the CSC presented at the 21 August 2014 workshop. These are shown relative to the SDOOL discharge volumes as a percent.
SCENARIO 1 m/day (4.8 ML/day) 2 m/day (9.6 ML/day)
VOLUME (GL/yr) PERCENT
2013 PERCENT 2032 VOLUME (GL/yr) PERCENT 2013 PERCENT 2032 1 site (individual) 1.7 3.4 2.1 3.5 6.9 4.3 2 sites (paired) 3.5 6.9 4.3 7.0 13.7 8.6 5 sites (inland) 8.7 17.1 10.7 17.5 34.3 21.4 6 sites (coastal) 10.5 20.6 12.9 20.9 41.2 25.7 11 sites (all sites) 19.2 37.7 23.6 38.4 75.5 47.1
The model results showed changes in groundwater levels relative to a ‘Business As Usual’ (BAU) case which may include a gradual fall in regional levels as the climate becomes slightly drier. The BAU assumes the current operation of the KWWTP infiltrating at 4.8 ML/d. An example of the effect of adding 4.8 and 9.6ML/d (1.7 and 3.5 GL/yr) at site 2 on levels is shown in Figure 10.2 which is just east of the ALCOA Refinery. The rise in groundwater levels is asymmetric with the greatest rises being up-gradient (to the east of the infiltration point) because groundwater flow is restricted from discharging to Cockburn Sound. This is because the mounds acts like an underground dam with inflow water backing up behind the mound ‘wall’. At 9.6 ML/d the impact could extend as far east as Thomson Lake, a Ramsar-listed wetland, and The Spectacle lakes. MAR closer to these lakes (e.g. sites 7 and 9 respectively) would have much greater impacts of course. Details of all simulations can be found in Appendix A.
Figure 10.3 shows predicted spatial extents for the difference in hydraulic head (greater than 15 cm) between the MAR scenario and the BAU scenario in 2032. These simulations were run separately (i.e. only on site at a time) but have been superimposed here for comparison purposes.
The largest impacts of MAR are predicted for four of the inland sites, and in the north for the coastal sites, which reflects the relative transmissivities of the aquifer (it being much higher in the south west). Large transmissivities prevent a groundwater mound from forming due to the greater dissipation of water. These early screening runs used a version of the model that was later recalibrated so that impacts in the southwest are now projected to be slightly greater than in this version. However the relativities remained the same so the selection of sites was not affected by this later refinement.
Figure 10.2 Predicted changes in the watertable (WT in metres) at site 2 computed based on the difference in hydraulic heads for 2032 less those for the BAU scenario for 2032.
Figure 10.3 Predicted spatial extent for 2032 for the difference in hydraulic head greater than 15 cm between the MAR scenario and the BAU scenario. For clarity, model results are shown in two separate images. The sites shown in the images were modelled separated, but are plotted together here for comparison. There is no change for site 9 as it is BAU.
MAR at multiple sites leads to cumulative impacts and much greater mounding across a larger area as is shown by the combination of all inland sites at 4.8 ML/d (understanding that this is BAU for the central site 9, thus the limited groundwater response) and 9.6 ML/d (Figure 10.4). The 4.8 ML/d
would utilise about 17% of current available wastewater or 11% of the water available in 2032 (Table 10.1). It would raise groundwater levels in the inland chain of Beeliar lakes by more than 1m
compared with the BAU case. Such a scenario would however require wastewater to be pumped inland from the SDOOL (and/or East Rockingham WWTP) because there is insufficient water at the Kwinana WWTP.
Figure 10.4 Impact on groundwater levels of adding 4.8 ML/day in the four inland site (left) and at 9.6 ML/day (right) which increases MAR at the Kwinana WWTP from 4.7 to 9.6 ML/day.
An analysis of particle tracking pathways was conducted (Figure 10.5). The early version of the model predicts that constituent particles transported by advection in groundwater should reach the coast for the 6 sites coastal situated near the SDOOL by 2032 but none of the inland sites would reach Cockburn Sound. Later analyses using the improved groundwater model gave similar results but also showed that particles can be intercepted by large pumping bores and may therefore not travel far from the site of infiltration. The maximum aerial extent of particle pathways or spread in map view is influenced by variations in hydraulic conductivity. In areas of high hydraulic conductivity, a narrow spread of pathways develops. The model predicts limited up-gradient advective transport and travel distances (e.g. distances are usually less than a few hundred metres).
Figure 10.5 The maximum areal extent of particle tracking pathways computed by MODPATH over the 20 year period, 2013 to 2032. Coastal sites have a shorter travel time due to as particles are transported to Cockburn Sound. MAR scenarios for each of the 11 sites were simulated separately, but are presented here on combined images for comparison.
An analyses of the strengths and weaknesses of MAR in the eleven sites arising from the first workshop concluded:
1. Site 1 is too far north for industry to benefit and may cause problems for mobilising nitrogen plumes into Jervois Bay. It is close to the Beeliar Nature Reserve but adding 9.6 ML/d could marginally raise water levels in Thomson Lake
2. Site 2 is located where the salt water intrusion in most evident and would benefit the northern part of the industrial area (near ALCOA). Travel times to the Sound range
between 8 and 11 years depending on the addition rate. Projected groundwater level rises are highest at this site compared with the other coastal sites, potentially resulting in movement of the salt water wedge towards the coast.
3. Site 3 would benefit industry around the Kwinana Power Station and possibly reduce salt water intrusion to the north. Travel times are between 2 and 4 years. Projected
groundwater rises are much lower at this site compared with sites 1 and 2.
4. Site 4 would benefit industry; particle tracking shows it would extend under the BP refinery site. Travel times range between 5 and 7 years. Projected groundwater levels are lower again compared with Site 3.
5. Site 5 would benefit industry in the CSBP area although groundwater rises would be very modest given the high transmissivity of the aquifer in this area. Travel times range between 5 and 7 years.
6. Site 6 is too far south to benefit industry. Groundwater level rises are negligible and travel times range between 3 and 5 years. Water could be sourced from the East Rockingham WWTP which will use oxidative ditch methods which produce low nitrogen levels. Being a conservation reserve there is a requirement that Water Corporation not infiltrate treated
wastewater on site. This has resulted in disposal to the ocean via the SDOOL. Volumes in this WWTP will increase substantially in future.
7. Site 7 has the potential to raise groundwater levels in Lake Thomson and the related Beeliar chain of lakes. Over time it could also add groundwater to the Henderson / Jervois area. The nearest industry is Cockburn Cement which may benefit depending on their licenses. The water may travel 2.4 to 3 km from the site by 2032.
8. Site 8 would provide indirect benefit to the chains of lakes to its north and south, and delayed impact to industry in the northern KIC area. The water may travel 1.9 to 2.6 km from the site. This area had the greatest reduction in groundwater levels between 1990 and 2012 (possible because of horticulture).
9. Site 9 is the Kwinana WWTP site and it was assumed that 4.8 ML/d would continue until 2032. It is about 4.5 km from the coast and infiltrated water is calculated to travel between 3.2 and 4.2 km between 2012 and 2032. Past infiltrated water has therefore almost
certainly reached the KIA and Cockburn Sound having started in about 1975 and travel times not being very sensitive to volumes. Doubling infiltration at this site (if water was diverted from the SDOOL) would raise levels in The Spectacles by 0.5 to 1 m. It would also raise levels in a wooded depression immediately east of the DAFWA Medina Research Station. Its effect on site 3, if jointly operated, is to increase water for industry in the central KIA area.
10. Site 10 is near the Kwinana townsite and therefore close to residential and council bores. It is not close to industry or important wetlands. Travel distances range between 2.8 and 3.4 km over the 20 year period.
11.
Site 11 is near Bollard Bullrush Swamp. Added water is likely to move south west towardsLake Cooloongup which is naturally saline.
It was therefore decided to concentrate most effort on sites 2, 3, 4, 5, 7 and 9.