subsidence modelling related to squeeze mining phase 1a ... · sgs subsurface consultancy 15...
TRANSCRIPT
Subsidence modelling related to Squeeze
Mining – Phase 1a results (v2017)
June 2017
2 SGS Subsurface Consultancy
Presentation outline
Introduction subsidence modelling Phase 1a
Input data comparison v2016 vs v2017
Subsidence modelling
Simplified model (squeeze volume in TR-1 location)
Detailed history match (squeeze volume is allocated between
cavern locations)
Impact on forecast
Conclusions
3 SGS Subsurface Consultancy
Introduction: Squeeze modelling Phase 1a
Objective
History match the historical subsidence due to Nedmag squeeze
mining for the time period 1993-2016 using the Geertsma – van
Opstal approach with variable rigid basement
– Attempt allocation of squeeze volumes to specific wells
Data available
X, Y, Z cavern locations for 13 existing wells
Nedmag-calculated cumulative squeeze volumes
– Per individual well
– Per cluster after cavern connection
Nedmag-interpreted cavern connection times
Antea-processed subsidence at benchmark points for 1993-2016 due
to salt squeeze mining (processed from the original data by
‘objectpunt’ analysis to separate subsidence from different sources)
– 115 initial subsidence benchmark points, available from 1993
– 264 additional points that were added to the network later
4 SGS Subsurface Consultancy
Introduction: Phase 1a – original and v2017
The original Phase 1a was carried out in 2016
Conclusion: Historical squeeze volumes and subsidence
benchmark data not in line with each other
Mismatch between squeeze volume and subsidence data ~ 40%
(squeeze volume > subsidence volume)
Benchmark dataset used not properly corrected for reference point
subsidence
Phase 1a was re-done in 2017 (Phase 1a v2017) with modified
input data – focus of this presentation
Nedmag revised their squeeze volume calculations
Antea / Nedmag re-processed the benchmark subsidence dataset to
include reference point subsidence
5 SGS Subsurface Consultancy
Input data Phase 1a v2017
Data available Phase 1a v2017, changes with respect to original
Phase 1a indicated:
X, Y, Z cavern locations for the existing 13 wells
Revised Nedmag-calculated cumulative squeeze volumes
– Per individual well
– Per cluster after cavern connection
Nedmag-interpreted cavern connection times
Revised, Nedmag-processed subsidence at benchmark points due to
salt squeeze mining (processed from the original data by ‘objectpunt’
analysis)
– 115 initial subsidence benchmark points, available from 1993
– 264 additional points that were added to the network later*
* Available if required, currently not in use because modelling is performed based on cumulative subsidence, while these points
represent differential subsidence. Cumulative subsidence was chosen for modelling to reduce the influence of noise / inaccuracies
in the dataset (such as points showing uplift in certain time intervals) and for numerical efficiency.
6 SGS Subsurface Consultancy
Input data: Squeeze volume calculation
The 2017 revised squeeze volumes are the result of new mass
balance calculations carried out by Nedmag incorporating the
expansive volumetric effects of:
direct dissolution of Bischofite and
thermal expansion of the underground brine
Version 2 of the Nedmag mass balance study has been
submitted to SodM on February 27th 2017
7 SGS Subsurface Consultancy
Input data – original vs v2017: Squeeze volumes
Comparison of Nedmag-provided squeeze volumes, original
(v2016) vs revised (v2017)
0.0
1.0
2.0
3.0
4.0
5.0
6.0
19
93
19
94
19
95
19
96
19
97
19
98
19
99
20
00
20
01
20
02
20
03
20
04
20
05
20
06
20
07
20
08
20
09
20
10
20
11
20
12
20
13
20
14
20
15
20
16
20
17
Sq
ue
eze
vo
lum
e,
mln
sm
3
Date
v2017
v2016
Squeeze volume @ March
2016 is ~ 13% smaller in v2017
8 SGS Subsurface Consultancy
Input data: Benchmark data correction
The subsidence data
were corrected by
Nedmag for AESubs
calculated subsidence of
the 12 reference
benchmarks using the
‘objectpunt analyse’
methodology
9 SGS Subsurface Consultancy
Input data – original vs v2017: Benchmark data
Comparison of benchmark
datasets, original (v2016) vs
revised (v2017) – initial points
only
Note: In v2017 dataset, corrected
benchmark data are available for
“initial” points only
v2017 vs v2016 observations:
Smaller scatter in v2017 data at
distances larger than 2000 m
from the TR-1 cavern center
Larger subsidence in v2017 data
in the flank areas
v2017
v2016
Su
bsid
en
ce fro
m b
en
ch
mark
da
ta,
in m
m
Su
bsid
en
ce fro
m b
en
ch
mark
da
ta,
in m
m
10 SGS Subsurface Consultancy
Input data – v2017: Benchmark data (I)
Two squeeze “centers” can be
recognized by visual
inspection of the cumulative
subsidence from benchmark
data on 04-2016:
TR-1 (TR-2) location
VE-1 (VE-2) location
Main subsidence is
approximately above TR-1
(TR2) location
v2017
Subsidence data at benchmark locations
interpolated with cubic interpolation
Subsidence from benchmark data
on 04-2016 (in mm)
11 SGS Subsurface Consultancy
Subsidence modelling
Subsidence modelling in order to obtain a history match for the
time period 1993-2016 was carried out in two steps (cf. original
Phase 1a in 2016):
Simplified method: To obtain rough insight into match between total
squeeze volume and subsidence bowl volume
– Assign full squeeze volume to a single location (TR-1 cavern centre)
Detailed history match
– Allocation of squeeze volumes between the various wells / clusters
12 SGS Subsurface Consultancy
Simplified model – model specifications
Input data
One point source location – “TR-1” cavern center
Fitting data
Cumulative subsidence in the “initial” benchmark point locations from
January 1999 to March 2016.
– Data between 1993 and 1999 show scatter near the center of the bowl
and was not used
Calibration parameters and constraints
Cumulative squeeze volume (Range: 0.0 – 5.6e+6 Sm3)
Rigid basement depth (Range: 1800 – 10000 m)
Calibration was performed independently for every subsidence
measurement time step
Objective function
Root mean squared error RMSE
Minimization method – Powell algorithm
13 SGS Subsurface Consultancy
Summary of optimisation runs
The best parameter set has shifted when using the v2017 data:
The larger cumulative squeeze volume is in line with the observation that the
subsidence in larger in the flank area in v2017 data set
Deeper rigid basement depth
Sum of
errors
(RMSE)
Sum of
errors
(RMSE)
Maps indicate single global minimum
Error density map, final time step v2017 Error density map, final time step v2016
14 SGS Subsurface Consultancy
Quality of history match with simplified method
Quality of the match is good
Same for v2016 and v2017
In v2017 less scatter at the distance >2000 m from the TR-1 cavern center
v2017
Benchmark subsidence
Modelled subsidence
v2016
Benchmark subsidence
Modelled subsidence
15 SGS Subsurface Consultancy
Modelling results: Match with Nedmag provided squeeze volume
Mismatch between modelled and Nedmag-calculated historical squeeze volume is ~10%
Results of sensitivity study on the subsidence bowl volume are in line with Nedmag conclusions (4.33
mln. sm3 vs 4.49±0.26 mln sm3)
Date v2016 v2017
Nedmag provided
squeeze volume, sm3
Modelled squeeze
volume, sm3
Modelled / Nedmag prov.
squeeze volume, %
Nedmag provided
squeeze volume, sm3
Modelled squeeze volume,
sm3*
Modelled / Nedmag prov.
squeeze volume, %
Mar-16 5686150 3668118 64.5 4925577 4331444 87.9
Feb-14 5137462 3302366 64.3 4503023 3989265 88.6
Apr-12 4670844 2948155 63.1 4095653 3639670 88.9
Jan-10 3964964 2514562 63.4 3505141 3038207 86.7
Jan-08 3505762 2180258 62.2 3122759 2848164 91.2
Jan-06 3010418 1861517 61.8 2724744 2353722 86.4
Jan-04 2510530 1556660 62.0 2297674 2156421 93.9
Jan-02 1988924 1229853 61.8 1853106 1667802 90.0
Jan-00 1538168 935058 60.8 1456876 1443921 99.1
Jan-99 1291992 765600 59.3 1237705 1194395 96.5
Nedmag conclusions for v2017 dataset * sm3 = m3 at surface conditions
16 SGS Subsurface Consultancy
Modelling results: Squeeze volumes v2016 vs v2017
Modelled
cumulative
squeeze
volume for
v2017
subsidence
data is larger
than v2016
0.0E+00
5.0E+05
1.0E+06
1.5E+06
2.0E+06
2.5E+06
3.0E+06
3.5E+06
4.0E+06
4.5E+06
5.0E+06
19
95-
01-
01
19
96-
01-
01
19
97-
01-
01
19
98-
01-
01
19
99-
01-
01
20
00-
01-
01
20
01-
01-
01
20
02-
01-
01
20
03-
01-
01
20
04-
01-
01
20
05-
01-
01
20
06-
01-
01
20
07-
01-
01
20
08-
01-
01
20
09-
01-
01
20
10-
01-
01
20
11-
01-
01
20
12-
01-
01
20
13-
01-
01
20
14-
01-
01
20
15-
01-
01
20
16-
01-
01
Ca
lcu
late
d c
um
. sq
ue
eze v
olu
me
in T
R-1
lo
ca
tio
n,
mln
sm
3
v2017
v2016
Linear (v2017)
Linear (v2016)
17 SGS Subsurface Consultancy
Detailed history match – model specifications
Input data
Multiple source points locations in cavern centers
– With time, relevant wells are connected / added into a single cluster based on Nedmag
provided cavern connection times
100% of Nedmag calculated squeeze volume
Fitting data
Cumulative subsidence at “initial” benchmark points locations from January
1993 to March 2016
Calibration parameters and constraints
Cluster production allocation fractions
Rigid basement depth (modified TNO-AGE approach)
Calibration was performed simultaneously for all subsidence
measurement time steps
Objective function
Normalized mean squared error
Minimization method – SLSQP algorithm
18 SGS Subsurface Consultancy
Summary of optimisation runs
25 simulation runs with random initial
parameters were performed to explore
an uncertainty space
Two local minima can be identified
(highlighted in the table in green and
red color)
Parameter range
c/k(0) in a narrow range 0.701 – 0.707
D(c/k) in a wide range 0.502 – 1.0
Tau in a wide range 33.3 – 78.2 year-1
# D(c/k) c/k(0) tau Error* 22 0.502 0.707 33.3 74254.7 7 0.578 0.705 39.9 74314.5
18 0.569 0.706 39.1 74322.3 3 0.566 0.706 38.8 74322.4
21 0.568 0.706 39.0 74324.1 5 0.565 0.706 38.7 74327.1 8 0.632 0.705 44.6 74378.3
19 0.636 0.705 44.8 74383.1 13 0.634 0.705 44.7 74388.0 0 0.675 0.704 48.3 74416.8
15 0.720 0.704 52.2 74445.0 12 0.826 0.703 61.3 74504.1 14 0.830 0.703 61.7 74509.6 11 0.852 0.703 63.6 74511.5 24 0.972 0.702 74.0 74559.9 2 1.000 0.701 78.2 74615.2
16 0.006 0.001 58.8 135858.7 9 0.001 0.001 56.3 135859.0
23 0.007 0.001 77.6 135859.1 20 0.008 0.001 84.1 135859.2 1 0.003 0.001 30.4 135859.4 6 0.007 0.001 69.1 135859.4 4 0.006 0.001 59.4 135859.8
10 0.003 0.001 26.0 135859.9 17 0.003 0.001 22.8 135860.2
* sum of errors for all subsidence surveys
#22
: B
est re
sult
(sm
alle
st e
rro
r)
19 SGS Subsurface Consultancy
Observed vs calculated subsidence (I)
Subsidence
simulation run #22
based on following
parameters:
c/k(0) = 0.707
D(c/k) = 0.502
tau = 33.3 year-1
“Typical” results of subsidence simulation (for runs with green colour on the previous slide)
Run#22 (=run with smallest error)
Subsidence from benchmark data
on 04-2016 (in mm) Modelled subsidence
on 04-2016 (in mm)
20 SGS Subsurface Consultancy
Observed vs calculated subsidence (II)
Some mismatch in the modelled subsidence in VE cavern center (underprediction)
and in the flank area (overprediction)
Note the difference map is affected by interpolation of the benchmark points
Run#22
Difference Benchmark – Modelled
subsidence on 04-2016 (in mm)
Benchmark subsidence (in mm)
Mo
de
lled s
ub
sid
ence (
in m
m)
Mo
de
l u
nd
ere
stim
ate
s
Mo
de
l o
ve
restim
ate
s
21 SGS Subsurface Consultancy
Observed vs calculated subsidence (III)
Difference benchmark subsidence vs modelled subsidence, interpolated with
cubic interpolation (left) and non-interpolated as bubble map (right)
Run#22
Difference Benchmark – Modelled
subsidence on 04-2016 (in mm)
Mo
de
l u
nd
ere
stim
ate
s
Mo
de
l o
ve
restim
ate
s
Difference Benchmark – Modelled
subsidence on 04-2016 (in mm)
underprediction
overprediction
22 SGS Subsurface Consultancy
Difference maps @ Apr 2016
Difference maps between benchmark subsidence and modelled subsidence
are very similar for all runs
Run#22 Run#2 Run#15
Difference Benchmark – Modelled
subsidence on 04-2016 (in mm) Difference Benchmark – Modelled
subsidence on 04-2016 (in mm)
Difference Benchmark – Modelled
subsidence on 04-2016 (in mm)
23 SGS Subsurface Consultancy
Subsidence at benchmark location: Benchmark data
vs modelled subsidence - examples
Three benchmark
points (115, 132, 136)
at different distance
from TR-1 cavern
center
24 SGS Subsurface Consultancy
Subsidence in point #115 (near TR-1)
Run#22
Modelled subsidence
Benchmark subsidence
incl. error bar (± 10 mm)
25 SGS Subsurface Consultancy
Subsidence in Point # 136 (mid-distance)
Run#22
Modelled subsidence
Benchmark subsidence
incl. error bar (± 10 mm)
26 SGS Subsurface Consultancy
Subsidence in Point # 132 (subsidence bowl rim)
Run#22
Modelled subsidence
Benchmark subsidence
incl. error bar (± 10 mm)
27 SGS Subsurface Consultancy
Benchmark vs modelled subsidence – same scale
Run#22
Point 115 Point 136 Point 132
Modelled subsidence
Benchmark subsidence
incl. error bar (± 10 mm)
28 SGS Subsurface Consultancy
Squeeze volume allocation
The largest part of squeeze volume is associated
with TR-1 cavern location (>70%)
Results of the model calibration are in line with visual
inspection of subsidence data on slide#7 (bowl
center in TR-1 (TR-2) location)
Cavern
Fraction of
cumulative cluster
squeeze volume (Jan
1997 – Apr 2016), %
TR-1 71.3%
TR-2 19.7%
TR-3 3.2%
TR-5 5.8%
Well squeeze volume fraction in the total cluster squeeze volume (Run#22)
DATE Cum. production from
clusters, mln.m3
Allocation well fraction
TR-1 TR-2 TR-3 TR-4 TR-5 TR-6 TR-7 TR-8 TR-9 VE-1 VE-2 VE-3 VE-4
Feb-1995 0.000 - - - - - - - - - - - - -
Jul-1995 0.000 - - - - - - - - - - - - -
Jan-1996 0.000 - - - - - - - - - - - - -
Jan-1997 0.017 0.00 1.00 - - 0.00 - - - - - - - -
Jan-1998 0.096 0.00 0.15 - - 0.85 - - - - - - - -
Jan-1999 0.240 0.00 0.00 - 0.00 1.00 0.00 - - - - - - -
Jan-2000 0.349 1.00 0.00 0.00 0.00 0.00 0.00 0.00 - - - - - -
Jan-2002 0.660 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 - - - - -
Jan-2004 1.089 0.73 0.00 0.27 0.00 0.00 0.00 0.00 0.00 - - - - -
Jan-2006 1.496 0.67 0.33 0.00 0.00 0.00 0.00 0.00 0.00 - - - - -
Jan-2008 1.874 0.80 0.20 0.00 0.00 0.00 0.00 0.00 0.00 - - 0.00 0.00 -
Jan-2010 2.257 0.37 0.63 0.00 0.00 0.00 0.00 0.00 0.00 - - 0.00 0.00 0.00
Mar-2012 2.847 0.61 0.39 0.00 0.00 0.00 0.00 0.00 0.00 - - 0.00 0.00 0.00
Feb-2014 3.246 0.99 0.01 0.00 0.00 0.00 0.00 0.00 0.00 - - 0.00 0.00 0.00
Apr-2016 3.625 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 - - 0.00 0.00 0.00
29 SGS Subsurface Consultancy
Detailed history match vs. simplified model results comparison
Detailed history match
simulation results are in
better agreement with
benchmark data (with
few exceptions only, but
the difference is small)
Date
Cumulative error (RMSE)
Detailed history
matching (Run#22) Simplified model
Mar-16 0.1449 0.1621
Feb-14 0.1239 0.1483
Apr-12 0.1099 0.1308
Jan-10 0.0997 0.1109
Jan-08 0.0913 0.0945
Jan-06 0.0870 0.0836
Jan-04 0.0764 0.0727
Jan-02 0.0692 0.0661
Jan-00 0.0601 0.0726
30 SGS Subsurface Consultancy
Impact of parameter range on forecast (assuming
subsidence from the single cavern center)
Model calibration Forecast
Minor
difference
Larger
difference in
the forecast
31 SGS Subsurface Consultancy
Conclusions
Good quality history match was obtained for the new
squeeze volume & subsidence data set
Results of sensitivity study are in line with Nedmag
conclusions. The calculated volume of the subsidence bowl is
within Nedmag’s confidence interval
Several parameter sets providing very similar history
match quality and different rigid basement forecast
behaviours were identified
Propose to use scenarios #22 and #2 in the forecast (dark
blue lines on previous slide)