Numerical Simulation of Basal Aquifer Depressurization in the … · 2016-01-22 · Numerical...
Transcript of Numerical Simulation of Basal Aquifer Depressurization in the … · 2016-01-22 · Numerical...
Numerical Simulation of Basal Aquifer
Depressurization in the Presence of
Dissolved Gas
Karl P. Lawrence, Ph.D.
Rock Engineering Division
Golder Associates
Mississauga, Ontario, Canada
Outline
Motivation:
Walday Abeda, Canadian Natural
Background:
The importance of basal aquifer depressurization in oil sand mining and
the issue of dissolved gas
Problem Statement
Approach:
Sampling for Dissolved Gas;
Groundwater (GW) Modelling
Results & Conclusions
April 17, 2012 2
Motivation
Walday Abeda (Canadian Natural)
April 17, 2012 3
Background Basal Aquifer Depressurization and Dissolved Gas
April 17, 2012 4
Background - Hydrogeology
April 17, 2012 5
Oil Sand
Water Sand
Background – Aquifer Isopach
April 17, 2012 6
Basal Aquifer
thickness and mine
footprint
Background – Standard Modelling
General Modelling Process:
Construct 3D models to
simulate static groundwater
conditions and
transient flow
Calibrated to static/
pre-mining conditions
Predictive simulations to determine
necessary flow rates/volumes/schedules and number/location of DP wells
April 17, 2012 7
Background – Challenge of Gas
DP modelling is typically performed assuming single-phase (water) flow.
The concentration and composition of gas dissolved in the groundwater
may cause near well “gas-locking” during pressure reduction:
DP wells depressurize aquifer;
Pore pressure reductions lower the gas solubility;
When pressure reduces below critical value (“bubble pressure point”),
exsolution occurs; and
The effects include increase in effective specific storage or decrease in
effective permeability.
April 17, 2012 8
Background – Challenge of Gas
April 17, 2012 9
PW-A OW-A
THEORETICAL
DRAWDOWN CONE PW-A OW-A
BUBBLE PRESSURE
POINT
NO GAS WITH GAS
Background - Gas Locking Indicators
April 17, 2012 10
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
0 30 60 90 120 150 180 210 240 270 300 330 360 390 420 450 480 510 540 570 600 630 660 690 720 750 780 810 840 870 900
Elapsed Time (min)
Dra
wd
ow
n (
m)
DP-06 Datalogger OBW @ DP-06 Datalogger OBW @ DP-06 Manual DP-06 Manual
Canadian Natural Horizon Oil Sands Project
2007 Basal McMurray Watersands Drilling and Testing Program
DP-06 Arithmetic Plot
Prepared By: Prepared For:Figure No.
Separation between pumping well and observation well = 31.0 m
Q1 = 157 m3/d (24 igpm)
Q2 = 255 m3/d (39 gpm)
Q3 = 360 m3/d (55 igpm)
Q1
Q2
Q3
22
Available drawdown to pump intake = 65.6m
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
0 30 60 90 120 150 180 210 240 270 300 330 360 390 420 450 480 510 540 570 600 630 660 690 720 750 780 810 840
Elapsed Time (min)
Dra
wd
ow
n (
m)
DP-03 Datalogger OBW @ DP-03 Datalogger OBW @ DP-03 Manual DP-03 Manual
Canadian Natural Horizon Oil Sands Project
2007 Basal McMurray Watersands Drilling and Testing Program
DP-03 Arithmetic Plot
Prepared By: Prepared For: Figure No.
Separation between pumping well and observation well = 31.0 m
Q1 = 157 m3/d (24 igpm)
Q2 = 333 m3/d (51 gpm)
Q3 = 655 m3/d (100 igpm)
Q4 = 982 m3/d (150 igpm)
Available drawdown to pump intake = 65.1m
Q1
Q2
Q3
Q4
13
No gas locking
DP well appeared to become gas locked
1) Excessive gas noted in discharge
water.
2) Inflection point in drawdown curve
indicative of change in “effective”
permeability
Background - Effect of Gas Locking
DP wells have a reduced efficiency and area of influence; DP is locally
limited;
Additional DP wells may be required to achieve target water level
elevations;
Failure to identify dissolved gases can result in:
1. Significant scheduling delays
2. Operational issues
3. Stability issues
4. And ….
April 17, 2012 11
Problem Statement
April 17, 2012 12
Problem Statement
Existing GW Flow Model and DP Well Network
If we do not account for dissolved gas in flow model predictions:
Discharge water volumes are over-estimated;
Rate of depressurization is over-estimated; and potentially
Number of required wells is under-estimated.
Primary objectives of numerical simulations:
1. Simulate the effects of gas-locking of DP-wells;
2. Design a DP well network capable of achieving target DP; and
3. Maintain safe workplace; save money; avoid scheduling delays.
April 17, 2012 13
Problem Statement
This is a reservoir engineering challenge applied to a large scale
groundwater flow modelling and mine planning scenario. Sampling
techniques to quantify gas were available from the oil industry but they
needed modification. Groundwater flow modelling needed to account for
variable permeability. A multi-disciplinary approach was required and a
meeting of minds to develop a management tool for Canadian Natural.
April 17, 2012 14
Approach
Characterization of Gas
April 17, 2012 15
Approach – Characterization of Gas
The primary milestones in developing a management tool were:
Characterization of dissolved gas through collection of pressurized
field samples and laboratory analysis;
Characterization of dissolved gas/effective permeability relationships
through multi-phase flow modelling to generate relative permeability
curves (Canadian Natural reservoir engineering); and
Incorporation of gas effects in groundwater flow modelling:
Standard approach if aquifer pressure > bubble point pressure
Modified approach with transient effective permeability when aquifer
pressure < bubble point pressure (based upon relative permeability
curves)
April 17, 2012 16
Approach – Characterization of Gas
AGAT Laboratory modified an oil
field bottom hole sampler for low
pressure conditions.
Sample of water collected under low
flow conditions at ambient aquifer
pressure adjacent to well screen.
Estimate saturation pressure (about
300 kPa) and gas composition.
Open top well makes
characterization approximate
however, consistency obtained
across site
April 17, 2012 17
Approach
Modified Groundwater Flow Model
April 17, 2012 18
Approach – Modified Model
Needed to modify existing groundwater flow model (MODFLOW) to
account for dissolved gas issues and transient effective permeability.
Modification involved:
Interruption of MODFLOW simulation at discrete time steps;
Automated update of hydraulic conductivity during discrete steps based
upon:
Current aquifer pressures; and
Permeability reduction relationship.
April 17, 2012 19
Approach – Modified Model
April 17, 2012 20
210
220
230
240
250
260
270
280
290
300
0 50 100 150 200 250 300 350
Co
rre
cte
d H
ead
(mas
l)
Time (min)
SIM-NO GAS LOCKING
DP-03-MEASURED
DP-03-SIM
DISCRETE STEPS
OBW4-MEASURED
OBW4-SIM
Approach – Modified Model
Relative permeability modification incorporated;
Steady-state calibration performed;
Transient calibration based upon:
Two DP well testing programs; and
History match performed based upon 1.5 years of historical pumping
data.
April 17, 2012 21
Time = 0 days
Vertical well
DP Well Rate
(m3/day)
DP-01 500
DP-02 50
DP-03 200
DP-04 250
DP-05 300
DP-06 300
02
01
03 04
05
06
Approach – Modified Model
Effect on K
Time = 5.3 days
Approach – Modified Model
Effect on K
Time = 9.4 days
Approach – Modified Model
Effect on K
Time = 13.5 days
Approach – Modified Model
Effect on K
Time = 17.6 days
Approach – Modified Model
Effect on K
Time = 23.9 days
Approach – Modified Model
Effect on K
Time = 30.3 days
Head minimum at wells
Approach – Modified Model
Effect on K
Approach – Modified Model
Effect on Depressurization
Permeability Relationship:
30% Reduction in Permeability / 10 m head reduction
Pump for 30 days:
Permeability reduction results in
steeper drawdown curve;
Radial extent of depressurization
reduced.
Example
Head at pump = 210 masl
Head at 8 m from well: 222 masl
Head at 25 m from well: 225 masl
225 230
220
225
Approach - Sample WL Contours
Vertical Wells
NO GAS LOCKING WITH GAS LOCKING
Head reduced to ~ 200 masl Head reduced to ~ 200 masl
at DP wells but only to ~ 230 masl
elsewhere
Approach - Sample WL Contours
Horizontal Wells
NO GAS LOCKING WITH GAS LOCKING
Head reduced to ~ 200 masl Improved performance over
vertical wells;
Results
Modified Groundwater Flow Model
April 17, 2012 32
Results – Model Sequencing
April 17, 2012 33
Area 1: History
Match
Area 6
Area 2
Area 3
Area 4 Area 5
Results
Modified Modflow approach
calibration:
Gas-locked DP well field and
simulated test showed an
excellent match.
Satisfactory match to complicated
DP well pumping history involving
33 scheduled stress (pumping)
periods; for
19 pumping wells.
April 17, 2012 34
-700
-600
-500
-400
-300
-200
-100
0
0 100 200 300 400 500 600 700 800 900 1000
Pu
mp
ing
Rat
e (
m3/
day
)
Time (days since startup)
DP-01 DP-02
DP-03 DP-04
DP-05 DP-06
DP-07 DP-08
DP-10 DP-12
DP-13 DP-14
DP-15 DP-16
DP-17 DP-18
DP-19 DP-20
DP-32 G10-044
G10-047 G10-049
G10-050 G10-058
G10-059
0
200
400
600
800
1000
1200
1400160
180
200
220
240
260
280
300
Rate
(m
3/d
ay)
Ele
vati
on
(m
)
South Pit Basal Aquifer Piezometric Heads
G10-041 G10-041-SIM
G10-042 G10-042-SIM
G10-043 G10-043-SIM
G10-045 G10-045-SIM
G10-057 G10-057-SIM
RATE - AREA 1 RATE - AREA 2
Results
Areas 3 and 4 were grouped due to aquifer connectivity
April 17, 2012 35
Existing/Planned Wells
Additional wells required
due to gas-locking
+
+
An additional 7 to 8
pumping wells were
required in each modelled
area to reach the DP target
Results
Sequential mining stages modelled;
Channelized aquifer was advantageous in that aquifer lobes could be
isolated (as a result of gas-locking).
We could use gas locking to our advantage to cut off the pressure
support to portions of aquifer requiring ongoing pit perimeter
depressurization;
Approximately twice as many wells were needed than originally planned;
and
Horizontal wells could potentially improve depressurization.
April 17, 2012 36
Conclusions
A pro-active approach was taken in addressing dissolved gas.
Other sites may experience the same issues in the future.
Substantial scheduling delays and
costs can be incurred if dissolved
gas is not identified and considered
in the pre-construction engineering
analysis.
Characterization of dissolved gas
is key!
April 17, 2012 37
Future (On-going) work
Discrete nature of MODFLOW modification presents subjective
interpretations;
Options:
1. Modify MODFLOW source code – achievable but pre/post-processing
issues arise;
2. Port modelling to FEFLOW – additional modules are permitted through
IFM to induce continuous effective permeability reduction in time
(without interruption to the simulation)
Option 2 is the focus of current work.
April 17, 2012 38
Acknowledgements
• Canadian Natural Horizon
• Co-authors:
Ken Baxter, Golder Associates
Walday Abeda, Canadian Natural
Walter Alexandru, Canadian Natural