Introduction. In-situ Steam Assisted Gravity Drainage (SAGD) oil sand operations require a source of fresh water for steam injection.

Transcription

1 Assessing Cumulative Effects of SAGD Operations in the Mackay Watershed Dirk Kassenaar, Earthfx Inc. Watertech 2016 April 7, 2015

2 Introduction In-situ Steam Assisted Gravity Drainage (SAGD) oil sand operations require a source of fresh water for steam injection. Groundwater supply wells, generally drawing from aquifers above the oil production zone, are a preferred source. From MEG Energy Corp. - Introduction 2

3 Study Objectives In 2014, Earthfx Inc. was hired by the Cumulative Environmental Management Association soc (CEMA) to answer the following question: Is there enough water in the Mackay watershed to sustain a responsible level of development Cumulative effects analysis requires the integrated assessment of: Multiple anthropogenic stresses: Numerous spatially distributed SW and GW diversions Land use change (land clearing, drill pads, roads, etc.) Intersecting effects on surface and groundwater systems: Changes in groundwater levels (drawdowns) in all aquifer systems Changes to frequency, duration and severity of low flow conditions - Introduction 3

4 Study Area Legend Lake Namur Lake MacKay River Watershed is located immediately north-west of Fort McMurrray, AB Includes Syncrude Mine site and numerous SAGD operations Watershed Area: 5,600 km 2 Model Area: 7,900 km 2 4

5 Study Approach Step 1: Integrated Model Development and Calibration Model Development: Compile Geology, Hydrogeology, Climate, Hydrology, Hydraulics Model Calibration: Build and pre-calibrate the SW and GW submodels Complete the fully integrated model calibration: Full reconciliation of entire hydrologic cycle water budget Step 2: Sustainability Assessment Define Assessment Criteria and Climate Period Define aquifer drawdown and streamflow impact sustainability thresholds Select a representative surrogate climate assessment period (25 years) Simulate Pre-development (Baseline), Current and Full Build conditions over the climate period Compare, on a daily basis, Current and Full-build conditions against Baseline Evaluate GW drawdowns and streamflow changes against Assessment Criteria - Introduction 5

6 MODELLING APPROACH - Modelling Approach 6

7 Integrated Modelling Approach: Advantages Study Approach: Fully integrated surface water and groundwater model Better representation of: Groundwater recharge and Dunnian GW feedback Streamflow and induced leakage SW/GW storage Cumulative effects of all SW and GW diversions Flux inputs and calibration targets Measured precipitation as input Calibration to total streamflow and measured GW levels - Modelling Approach 7

8 Selected Model: USGS GSFLOW USGS integrated GW/SW model Based on MODFLOW-NWT and PRMS (Precipitation-Runoff Modelling System) Open-source, proven and very well documented Fully-distributed: Cell-based representation Excellent balance of hydrology, hydraulics and GW - Modelling Approach 8

9 SUMMARY OF MACKAY MODEL DEVELOPMENT 9

10 Study Area Features Topography (600 m of relief) Birch Mountains Thickwood Hills Incised river and stream network MacKay River main channel Dover and Dunkirk Tributaries Athabasca River: South and eastern boundary Legend and Namur Lakes Plus over 100 other lakes in study area Extensive muskeg and wetlands Bedrock Channel Aquifers Key GW supply source for multiple projects Anthropogenic Stresses Syncrude Mine SW and GW Diversions AMBI, 2013) 10

11 GSFLOW: Multi ti-resolution Climate inputs ( 2.5 km cells) GSFLOW is unique in that the resolution of the model can be adjusted to match key features Surface Hydrology/Soil Zone ( 200x200 m cells) Stream Network Linear 1-D Channel segments (4000 km of streams represented, independent of grid resolution) Sub-surface Hydrogeologic Layers ( 13 layers of 400x400 m cells) 11

12 Model Grid Fully distributed model: Every cell has unique properties GW grid: 400 m by 400 m cells Selected to match assessment averaging criteria (impact at 150 m from a well) but avoid focus on specific water users. Can be refined for future studies SW Grid: 200x200 m cells Improved representation of overland flow, wetlands, interflow and soil zone processes and properties Stream routing: All streams and rivers simulated 400x400 m GW grid 12

13 Geologic Setting Surficial Geology Bedrock Geology Predominantly tills and glaciolacustrine deposits Subcrop of units that dip to the southwest 13

14 Geologic Information Primary sources for geologic borehole data: 25,000 - Alberta Geological Survey Atlas (Western Canada Sedimentary Basin) Limited geologic data in Birch Mountains and central portion of study area 14

15 Conceptual Stratigraphic Model After AGS Source: Andriashek and Atkinson, 2007 Empress Channel Sands: Key water supply aquifer 15

16 Hydrostrat. Layers 19 layer strat. model used to produce 17 layer hydrostrat. model. Some units of similar hydraulic properties were combined. McMurray Basal Sands added as separate aquifer unit. Model does not extend below Prairie Fm. Aquiclude Assumed minimal communication due to low vertical K of unit. Simulating higher salinities (>50,000 mg/l TDS) would require density dependent groundwater flow. Period Quaternary Cretaceous Devonian Unit Stratigraphic Model Unit Hydrostratigraphic Model 1 Late Lacustrine 1 Aquitard 2 Surface Sand 2 Aquifer 3 Grand Centre Till 3 Aquitard 4 Middle Sands 4 Aquifer 5 Intermediate Till 5 Aquitard 6 Empress Channel Sands 6 Aquifer 7 Labiche Formation 7 Aquitard 8 Pelican/Viking Formation 8 Aquifer 9 Joli Fou Formation 9 Aquitard 10 Grand Rapids Formation 10 Aquifer 11 Clearwater Formation 12 Wabiskaw Formation 11 Aquitard 13 McMurray Formation 12 Aquitard (includes Basal Sands) 13 Basal Sand Aquifer 14 Winterburn Formation (not included) 15 Grosmont Formation 14 Aquifer 16 Lower Ireton Formation 15 Aquitard 17 Cooking Lake Formation 16 Aquifer 18 Beaverhill Lake Group 17 Aquitard 19 Watt Mountain Formation (Top of Elk Point Group) Base of Model Prairie Formation Keg River Formation Contact Rapids Formation (not included) Laloche Formation Precambrian 16

17 GW Level Data 803 wells with water level data Well assigned to hydrostrat. units based on screened intervals. Limited long term temporal monitoring data (GOWN) 17

18 Groundwater Submodel Calibration Steady-state submodel calibration. Better calibration in aquifers where data more plentiful Overburden Empress Fm. Labiche Aquitard Viking Aquifer Joli Fou Aquitard Grand Rapids Aquifer Simulated (masl) Clearwater Aquitard McMurray Aquifer/Aquitard Cooking Lake Aquifer 1:1 Error Intervals (±10 m) Unit Number of Wells (n) ME (m) MAE (m) RMSE (m) Range in Observations (m) RMSE as Percent of Range (%) Overburden % Empress % Labiche % Viking % Joli Fou % Grand Rapids % Clearwater % McMurray % Cooking Lake % Overall % Observed (masl) 18

19 Hydrologic Submodel Development (PRMS) Based on the USGS Precipitation-Runoff Modeling System (PRMS) code In each unique cell: Fully distributed implementationn 200m x 200m cells (196,832 unique cell HRUs) 19

20 Climate Inputs Precipitation and temperature interpolated on a daily basis over a 2.5km x 2.5km grid Inverse distance squared weighting 25 year daily climate time series input for each grid cell 20

21 Vegetative Cover Classes 26 wetland and vegetative cover classes 17 types of wetlands Model parameters assigned by class: Seasonal Cover density Vegetation indices Soil zone properties Overland flow and shallow interflow parameters 21

22 Overland Flow Overland flow and interflow simulated with a topographically defined cascade network 200x200m cell representation 22

23 Lateral Flow Processes PRMS Soil Zone Head Dependant Saturation Dependant MODFLOW Layer 1 MODFLOW Layer 2 23

24 Dunnian Flow Processes: SW/GW Feedback Groundwater feedback dominates in discharge areas, wetlands and shallow aquifers GW feedback in up to 60% of the watershed Complex transient runoff and rejected recharge Occurs when the water table is at or near surface Spatially controlled: Tends to occur in stream valley areas Seasonally controlled: Tends to occur in spring when WT is high GW discharge to the soil zone can become interflow or overland flow Overland flow can re-infiltrate downslope: 3D recharge Unsaturated zone Gravity drainage Recharge Ground-water flow Stream 24

25 Frozen Ground New frozen ground module developed for this study GSFLOW is Open Source! Based on a modified form of the Stefan Equation Derived by the U.S. Army Corps of Engineers Model code follows Emerson (1994) X f = 86,400K f I f L + C T a + I f 2t 0.5 where X f = depth of frost K f = thermal conductivity I f = frost index degree days L = latent heat C = volumetric heat capcity T a = mean annual soil temperature t = duration of the freezing period 25

26 Frozen Ground Response Frozen soil dynamics affect both surface and subsurface processes: SW Runoff and Recharge: Enhanced runoff during spring freshet, no winter recharge GW Discharge: Significantly reduced winter discharge to streams and wetlands 26

27 Model Calibration and Validation Calibrated then verified against over 38 year period A range of hydroclimatic conditions simulated Validation Calibration 27

28 Model Calibration and Validation Hydrologic submodel and the final integrated model were calibrated against streamflow observations at 6 Water Survey (EC) and RAMP stations Historical observations at discontinued stations were an important source of insight 28

29 Model Calibration and Validation Good match to streamflow observations at study area gauges Daily Nash-Sutcliffe 0.65 Monthly Nash-Sutcliffe 0.75 Good match to validation period: Model has adequate predictive power 29

30 Distributed Results 30

31 Distributed Results (GSFLOW) 31

32 GSFLOW GW/SW Outputs April May 32

33 GW/SW Animation Animation shows spring melt and changes in GW levels and streamflow 33

34 GSFLOW Outputs Spring change in water levels and streamflow 34

35 GSFLOW Outputs Spring change in water levels and streamflow 35

36 GSFLOW GW/SW Water Budgets Significant inter-annual and seasonal storage effects Lake Seepage Stream Leakage Surface Leakage Recharge Wells Net Const. Head Net Storage Lake Seepage Stream Leakage Surf Leakage Recharge Wells Net Const. Head Net Storage Flows (mm/yr) 0-5 Flows (mm/month) Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Avg 36

37 Model Development Conclusions The Mackay GSFLOW model represents the complex transient surface and subsurface process and their interaction and feedback Extensive submodel development and integrated model calibration was undertaken to all available data Key aspect of the integrated model calibration: Directly measured flux input: Precipitation Directly observed calibration targets: Total measured streamflow and GW heads Overall water budget must balance no water is gained or lost 37

38 ASSESSMENT SCENARIOS: CRITERIA AND RESULTS - Assessment Scenarios 38

39 Diversion Scenarios Baseline: No pumping Current Conditions: 4 Operations including 11 pumped wells. Current Operations Current Operations Full-Build Conditions: 14 Operations including 42 pumped wells. Current Operations - Assessment Scenarios 39

40 Land Use Change Processing facilities and well pads Assumed to be 100m by 100m gravel pads spaced 500m on center Reduced ET, due to the loss of vegetation, increased runoff Full Build Scenario: Drill pads are estimated to cover 6% of the planned project areas; Roads, pipelines, and facilities cover another 4%. - Recommendations for Phase 3 40

41 Assessment Climate Period 25 year period includes a range of hydroclimatic conditions Includes both wet years (1997) and drought years ( , 2009 and 2011). 5 year spin-up period before start of assessment Surrogate Climate Period - Assessment Scenarios 41

42 GW Sustainability Assessment Criteria In summary, it was agreed that the sustainable drawdown is 50% of the available drawdown in a confined aquifer. Threshold selection based on the Alberta Environment Water Conservation and Allocation Guideline for Oilfield Injection (AENV, 2006) For unconfined aquifers, 66% of the average saturated thickness was used. Available drawdown based on average water level determined by 20 year baseline simulation. Assessment Process: all three scenarios run using the same climate inputs Only difference is diversions and land use change Daily outputs for every model cell and stream reach saved for comparison Drawdown calculation Alberta Desktop Assessment If, under Current or Full-build development conditions, drawdowns exceeded this threshold on any particular day in a 20 year assessment simulation, the cumulative diversion was considered locally unsustainable. - Assessment Scenarios 42

43 Overburden Impacts Overburden Aquifers Percent of Total Percent of Total Available Drawdown Layer 1 Drawdowns - Assessment Scenarios 43

44 Channel Aquifers Empress Formation Aquifer Percent of Total Percent of Total Available Drawdown Layer 4 Drawdowns - Assessment Scenarios 44

45 Confined Aquifers Viking/Pelican Aquifer Percent of Total Percent of Total Available Drawdown Layer 5 Drawdowns - Assessment Scenarios 45

46 Deep Aquifers Grand Rapids Aquifer Percent of Total Percent of Total Available Drawdown Layer 8 Drawdowns - Assessment Scenarios 46

47 GW Sustainability Assessment Cumulative GW drawdowns are significant, in particular in the lower highly confined aquifer units Offset by the fact that lower units have much greater available drawdown On a watershed scale, GW drawdowns appear to broadly stabilize within the 20 year period, suggesting sustainable water use Localized zones where drawdown exceed 50% of total available drawdown - Assessment Scenarios 47

48 SW Sustainability Assessment Criteria Alberta Desktop Method: Simulated frequency-duration relationship is calculated for every reach under baseline conditions The discharge that is exceeded 80% of the time is the ecosystem baseflow (EBF) component. ADM Criteria 1: No surface water diversions are allowed below the 80% EBF threshold No diversion allowed when flow is below the lowest flows that occur up to 20% of the time. ADM Criteria 2: Above the 80% EBF threshold, up to 15% of the available flow can be diverted. 20 year Baseline simulation used to determine weekly EBF threshold in every stream reach - Assessment Scenarios 48

49 SW Sustainability Assessment Criteria Frequency-duration relationship calculated in the watercourse in a natural state. EBF Weekly Threshold for Mackay River at Fort McKay: - Assessment Scenarios 49

50 SW Sustainability Assessment Threshold for Mackay River at Fort McKay shown ADM Criteria 1 - fails for select days, as shown in red ADM Criteria 2 - never more than 15% diverted Numerous other stream locations also assessed - Assessment Scenarios 50

51 Local SW Effects While the overall watershed passes the ADM criteria at the Mackay outfall point, local streams fail the 15% ADM criteria i.e. GW diversions locally induce leakage that exceeds 15% of the EBF (ecological baseflow) - Assessment Scenarios 51

52 Sustainability Assessment Conclusions In summary, the analysis indicates that projected water use in the study area is broadly sustainable, from both a groundwater and surface water aspect, on a watershed scale. This conclusion is supported by two findings: Results indicate that drawdowns do not, on a watershed scale, appear to grow over time Accumulated streamflow losses do not exceed the 15% ADM threshold along the main channel of the Mackay and Dover Rivers. The results do indicate, however, that under the full build scenario, cumulative groundwater diversions appeared to create unsustainable local impacts, as measured by both the groundwater drawdown and ADM thresholds. - Assessment Scenarios 52

53 Water Budget Comparisons Pre-development shows how wet and dry years replenish and deplete storage (royal blue) Full build scenario shows greater fluctuations in storage Simulated Inflows and Outflows by Water Year - Pre-Development Conditions Lake Seepage Stream Leakage Surface Leakage Recharge Wells Net Const. Head Net Storage Simulated Inflows and Outflows by Water Year - Full Build Conditions Lake Seepage Stream Leakage Surface Leakage Recharge Wells Net Const. Head Net Storage Flows (mm/yr) 0 Flows (mm/yr) Assessment Scenarios 53

54 Water Budget Comparisons Winter pumping depletes storage, replenished by April recharge. Baseline Conditions Full-Build Conditions - Assessment Scenarios 54

55 Other Insights Winter pumping under frozen ground conditions depletes shallow aquifer storage Baseflow discharge in May is reduced by 50% due to freshet replenishment of shallow aquifer storage. Understanding seasonal and inter-annual storage is essential Average Monthly Groundwater Discharge to Streams (mm/year) Pre-Development Full Build Average Monthly GW Average Monthly GW Discharge to Streams Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec - Assessment Scenarios 55

56 Overall Conclusions Detailed, fully integrated SW/GW modelling can provide significant insight into both cumulative effects and watershed function. Numerous applications Local impact assessment, water budgeting, climate change, drought assessment, eco-hydrology and water management. - Recommendations for Phase 3 56