Proceedings of the 1998 Conference on Hazardous Waste Research 32

SIMULATION OF GROUNDWATER FLUXES

DURING OPEN-PIT FILLING AND UNDER
STEADY STATE PIT LAKE CONDITIONS
D.B. Stone and R.C. Fontaine
Geomega, 2995 Baseline Road, Suite 202, Boulder, CO 80303, Phone. (303) 443-9117, Fax.
(303) 938-8123
A critical component in determining post-mining pit water quality is knowledge oI groundwater Iluxes,
both as the pit Iills and aIter steady state conditions have been reached. Simulations oI the Iilling oI an open pit
in Crescent Valley, Nevada, were made to generate inputs Ior pit lake chemistry predictions. The simulations
show that the water table recovery is most rapid immediately aIter pumping stops, when the hydraulic gradients
are steepest. The maximum lateral extent oI water table drawdown occurs several years aIter pumping stops
because water continues to be derived Irom storage as the pit Iills. Under steady state conditions, the lake stage
is lower than the elevation oI the water table in the pit area prior to mining, and groundwater Ilow is directed
toward the pit lake, because evaporation Irom the lake surIace causes it to act as a groundwater sink.
INTRODUCTION
The South Pipeline Project is a proposed expansion oI Cortez Gold Mine`s open-pit mining
operations in the southern part oI Crescent Valley, which is located in north central Nevada (Figure
1). This proposed expansion would create a second pit adjacent to the permitted pipeline pit. To
obtain approval Ior the project, a number oI Iederal, state, and local permits need to be secured.
One oI the elements oI the permitting process is an assessment oI the potential environmental
impacts to groundwater resources. To perIorm this assessment, a numerical groundwater Ilow
model was developed to integrate regional hydrogeologic conditions, recharge Irom inIiltration,
evapotranspiration, and stresses induced by the mine dewatering operations. The numerical code
used to simulate groundwater Ilow was an enhanced version oI the U.S. Geological Survey`s three-
dimensional, Iinite-diIIerence groundwater Ilow code MODFLOW (McDonald and Harbaugh,
1988).
A critical component in determining post-mining pit water quality is predicting groundwater
inIlows into the pit over time. The standard version oI MODFLOW is limited in its ability to simulate
pit lake Iormation. The new LAK2 package Ior MODFLOW was used in this study to overcome
these limitations. The LAK2 package was selected because it can calculate the transient stage oI a
pit lake as it Iills, as well as accounting Ior precipitation and evaporation at the lake surIace and
groundwater inIlows or outIlows across multiple model layers (Council, 1997).
This study is thought to be unique because it includes the simulation oI initial pit lake Iormation
in the adjacent pipeline pit concurrent with large-scale, open-pit dewatering, as well as the ultimate
development oI a pit lake in both pits aIter dewatering ceases.
ABSTRACT
Key words: groundwater flux, pit filling, pit lake formation, modeling
Proceedings of the 1998 Conference on Hazardous Waste Research 33
GROUNDWATER FLOW MODEL SETUP AND CALIBRATION
The regional model domain and grid used in this study are shown in Figure 1. The model
domain includes all oI Crescent Valley, which is roughly 50 miles long by 20 miles wide and has a
drainage area oI approximately 700 square miles. The model is bounded by mountain ranges to the
south, west, and east and by the Humboldt River to the north. To represent the steep gradients due
to dewatering, a cell size oI 200 x 200 x 200 It was used near the mine. Cell sizes increase toward
the model boundaries to a maximum size oI 10,000 It horizontally and 3,000 It vertically. The
boundaries are, Ior the most part, deIined as no-Ilow along the crests oI the mountain ranges.
Constant head boundaries were speciIied along the Humboldt River and at one location along the
western edge oI the model domain where a limited amount oI groundwater enters the basin due to
inIlow Irom an adjacent valley. Basin-wide groundwater recharge was estimated Iollowing the
method reported by Maxey and Eakin (1949). Evapotranspiration was deIined on the basis oI
available data and was correlated with the distribution oI phreatophytes in the center oI the valley.
Groundwater withdrawals Ior domestic, municipal, and agricultural usage were also included in the
groundwater Ilow model.
The model was divided into 12 horizontal layers to represent the vertical domain, extending
Irom 9,000 It above mean sea level to an elevation oI 5,000 It below mean sea level. This extent
was necessary to simulate Ilow in bedrock below the basin-Iill deposits, which are thought to attain
a maximum thickness oI approximately 10,000 It.
Modeled bedrock units include carbonate, siliceous, volcanic, and intrusive rocks. Basin-Iill
deposits were divided into younger and older basin-Iill units. The younger basin-Iill units include
alluvial Ians, landslides, stream Ilood plains, and playas. The older basin-Iill units consist oI
semiconsolidated deposits oI conglomerate, sandstone, siltstone, Ireshwater limestone, evaporite,
and interbedded volcanic rocks. Altogether, 22 separate hydrolithologic units were assigned in the
model.
Extensive Iaulting in the mountain ranges surrounding Crescent Valley plays an important role in
the groundwater Ilow system. Regional Iaults, and a number oI smaller Iaults that were interpreted to
act as partial barriers to groundwater Ilow in the immediate vicinity oI the mine, were simulated in
the model by using the Horizontal-Flow Barrier (HFB) package Ior MODFLOW.
The model was calibrated to both historical water-level measurements and to data collected
during the Iirst two years oI dewatering the pipeline pit. The calibration also included matching
estimated groundwater Iluxes at appropriate points along the model boundary.
SIMULATION OF PIT LAKE FILLING
Methodology
A three-dimensional representation oI the ultimate pit lake was developed on the basis oI the
200 x 200 x 200 It discretization oI the model grid in the area oI the pits. Model grid cells adjacent
Proceedings of the 1998 Conference on Hazardous Waste Research 34
to the exterior oI the lake were designated as 'lake cells, and were assigned hydrogeologic proper-
ties corresponding to the rock types that will ultimately be exposed in the pit walls. The deepest part
oI the lake is approximately 700 Ieet below the static pre-mining water table elevation. Overall, the
pit lake spans Iive layers in the model (Figure 2).
Interactions between groundwater, the pit lake, and the atmosphere were simulated with the
new LAK2 package Ior MODFLOW. The LAK2 package (1) takes into account precipitation into
and evaporation out oI the pit lake; (2) provides both horizontal and vertical groundwater Ilux
components through the lake cells, which are used as inputs to the geochemical models; (3) keeps
track oI lake stage and components oI the volumetric budget through time; and (4) simulates inter-
actions with surIace streams, although this Ieature was not used in the present study.
Output oI groundwater Ilux through each 200 x 200 It area oI the ultimate pit surIace was
generated on a monthly basis during the Iirst Iew years oI pit Iilling, when Ilow rates into the pit were
the greatest. The times between outputs were increased during the later part oI the simulation as the
Iluxes diminished.
RESULTS
Four diIIerent pit conIigurations were simulated to analyze the potential eIIects oI diIIerent
mining options. Figure 3a shows Scenario 1, where the South Pipeline pit is continuous with the
deeper pipeline pit. In Scenario 2 (Figure 3b), a portion oI the waste rock removed Irom the South
Pipeline pit is placed in the pipeline pit, yielding a lake oI a smaller volume. Under Scenario 3
(Figure 3c), the South Pipeline pit does not exist, and the lake Iorms only in the pipeline pit. Finally,
in Scenario 4 (Figure 3d), the entire pipeline pit is backIilled with waste rock, and the lake Iorms
only in the South Pipeline pit.
In the case oI Scenario 1 (Figure 3a), pit lake Iilling was initiated during the latter stages oI
dewatering when mining operations are Iocused on shallower deposits in the South Pipeline pit.
Figure 4 shows (1) the excavation schedule Ior both the pipeline and South Pipeline pits, (2) the
predicted dewatering rates, and (3) the calculated lake stage during the Iinal period oI dewatering.
The lake initially Iorms in the deepest part oI the pipeline pit and then gradually enters the South
Pipeline pit during the last two years oI dewatering. This simulation is thought to be unique in that
large-scale, open-pit dewatering and pit lake Iilling were simulated concurrently.
Figure 5 shows the water level in the pit Ior each conIiguration aIter dewatering ceases. Water
level recovery is most rapid immediately aIter pumping stops, when hydraulic gradients are the
steepest. The diIIerences in lake stage recovery between the various scenarios are due to the Iact
that the simulated lakes have diIIerent volumes and diIIerent dewatering and Iilling histories.
During dewatering, excess produced water will be returned to the groundwater basin via a
series oI inIiltration galleries. Simulation results show that the drawdown at the end oI dewatering
will be eIIectively constrained to the north and to the south by the planned inIiltration (Figure 6).
Proceedings of the 1998 Conference on Hazardous Waste Research 35
However, the drawdown cone will continue to expand laterally Ior some period aIter dewatering
ceases, because water continues to be derived Irom storage in the basin-Iill aquiIer as the pit lake
Iills. Figure 6 shows 10- and 100-Ioot drawdown contours at the end oI dewatering and at the time
oI maximum lateral extent oI the drawdown cone, which occurs approximately 20 years aIter
dewatering ceases.
Figure 7 shows a proIile oI the water table along the axis oI Crescent Valley at various times
during its recovery to equilibrium. The low, Ilat portions oI the curves correspond to the pit lake
surIace, which grows with time until the lake becomes Iull. Model simulations indicate that the Iinal
lake stage will be approximately 15 Ieet lower than the elevation oI the water table in the pit area
prior to mining, because the lake acts as a localized evaporative sink Ior groundwater. Furthermore,
as the lake reaches equilibrium, all oI the water Ilowing into the lake is removed by evaporation, so
there is no Ilow out oI the lake to groundwater.
DISCUSSION
Overall, the LAK2 package was Iound to be extremely useIul, robust, and relatively straight-
Iorward to apply to simulations oI pit lake Iilling. However, proper use oI the LAK2 package Ior
this type oI application requires careIul setup oI individual lake cells: e.g., using too coarse oI a grid
can adversely impact the calculation oI the transient lake stage, and a missing lake cell can lead to
errors in the volumetric budget. Vertical discretization is also important, especially at later times
when the rate oI change oI the lake stage is relatively slow. For the purpose oI geochemical model-
ing, Iiner temporal discretization oI output is required early on when Iluxes into the pit are changing
rapidly, to better characterize the volume oI water Ilushing oxidized wall rock.
SUMMARY AND CONCLUSIONS
The LAK2 package Ior MODFLOW was used to simulate pit lake Iilling under Iour diIIerent
pit conIigurations, corresponding to diIIerent mining options. This study is believed to be one oI the
only ones to simulate large-scale, open-pit dewatering and pit Iilling at the same time. Results oI the
simulations show that the transient pit lake stage is impacted by the conIiguration oI the pit and the
previous dewatering and Iilling history.
The methodology described in this paper Ior simulating pit Iilling is generally applicable to any
MODFLOW-based model oI open-pit dewatering and post-mining pit Iilling. The generated
groundwater Iluxes and lake-stage outputs are critical components Ior geochemical modeling oI
post-mining pit water quality.
ACKNOWLEDGEMENTS
The authors would like to acknowledge Cortez Gold Mines Ior granting us permission to
openly discuss these results.
Proceedings of the 1998 Conference on Hazardous Waste Research 36
REFERENCES
Council, G., 1997. Simulating Lake-Groundwater Interaction with MODFLOW. In: K.J. Hatcher
(Ed.), Proc. 1997 Georgia Water Resources ConIerence, Athens, Georgia, University oI
Georgia, pp. 457-462.
Maxey, G.B., and T.E. Eakin, 1949. Ground Water in the White River Valley, White Pine, Nye, and
Lincoln Counties, Nevada, State oI Nevada, OIIice oI the State Engineer, Water Resources
Bulletin No. 8.
McDonald, M.G., and A.W. Harbaugh, 1988. A Modular Three-Dimensional Finite-DiIIerence
Ground-Water Flow Model. U.S. Geological Survey Techniques oI Water-Resources
Investigations, Book 6.
Proceedings of the 1998 Conference on Hazardous Waste Research 37
Figure 1. Regional groundwater Ilow model grid and domain.
Proceedings of the 1998 Conference on Hazardous Waste Research 38
Figure 2. Lake cells by model layer (Scenario 1).
Proceedings of the 1998 Conference on Hazardous Waste Research 39
Figure 3. Pipeline - South Pipeline pit cross-sections.
Proceedings of the 1998 Conference on Hazardous Waste Research 40
Figure 4. Mine dewatering schedule, dewatering rates, and pit lake stage during dewatering.
Figure 5. Hydrographs oI lake stage during recovery.
Proceedings of the 1998 Conference on Hazardous Waste Research 41
Figure 6. Predicted water table drawdown in basin-Iill deposits.
Proceedings of the 1998 Conference on Hazardous Waste Research 42
4200
4300
4400
4500
4600
4700
4800
4900
5000
0 50000 100000 150000 200000
Distance along Profile, South to North (feet)
E
l
e
v
a
t
i
o
n

(
f
e
e
t

a
m
s
l
)
Pre-mining
1 Year Recovery
3 Years Recovery
10 Years Recovery
60 Years Recovery
250 Years Recovery
Figure 7. North/south proIile oI water table recovery aIter dewatering ceases, Scenario 1.
Proposed Pit
Crescent
Valley
Township
Humboldt
River