Transcript Slides

Analysis on Induced Seismicity
in the Netherlands
Rob van Eijs, Frans Mulders, Manuel Nepveu, Cor Kenter
Berend scheffers
Research by TNO, applied by the Dutch onshore operators
contents
• Observations on induced seismicity
• Mechanics of fault slip
• How to predict
– Numerical modelling
– Probabilistic modelling
– Correlation between field parameters
and seismicity
Induced seismic activity can
be divided in two categories
• Triggered. This group of earthquakes are
caused by tectonic stresses. They would
probably have occurred sooner or later,
but their time-space proximity to human
activity indicate antropogenic activity.
• Truly induced. This group of earthquakes
are purely antropogenic in that stress
buildup can be traced directly to human
activity.
Induced
earthquakes in
the Netherlands
(status January 2001)
Seismic cross section Groningen gas field
Evidence for induced earthquakes at reservoir levels
SW
NE
0
0.5 1
2
Depth
-0.0 km
3 km
-1.0 km
-2.0 km
Zechstein Salt
-2.5 km
Rotliegend
Source: NAM,
KNMI
Top Zechstein
-3.0 km
Top Rotliegend
5
contents
• Observations
• Mechanics of fault slip
• How to predict
– Numerical modelling
– Probabilistic modelling
– Correlation between field parameters
and seismicity
Main mechanisms of induced
fault slip
• Poroelastic stress development
• Associated with reservoir contraction
Adushkin et al. 2000
General concept of stress development and
fault reactivation within a gas reservoir
Reservoir contraction
Roest and Kuilman
contents
• Observations
• Mechanics of fault slip
• How to predict
– Numerical modelling
– Probabilistic modelling
– Correlation between field parameters
and seismicity
Bergermeer gasfield
Finite element modelling
extended
fault plane
transition zone
between fault and
rock
fine mesh around
reservoir
reservoir
150 m
DY=150 m
for q=90o
Mulders 2003
Relative Shear Displacement
on the fault plane (RSD)
RSD [cm]
RSD [cm]
RSD [cm]
RSD [cm]
5.0 10.0
5.0 10.0
5.0 10.0
5.0 10.0
C
D
B
A
S
Z
Y X
C’
D’
D
C
B’
A’
B
A
B’
A’
S
Z
Y
X Mulders
2003
D’
C’
RSD [cm]
12.0
10.0
8.0
6.0
4.0
2.0
0.0
Seismic moment
M0 = G · A · RSD
G = shear modulus
Aki (1966)
contents
• Observations
• Mechanics of fault slip
• How to predict
– Numerical modelling
– Probabilistic modelling
– Correlation between field parameters
and seismicity
Frequency-magnitude relation
KNMI 2002
contents
• Observations
• Mechanics of fault slip
• How to predict
– Deterministic
– Probabilistic
– Correlation between field parameters
and seismicity
Some other observations
• No earthquakes yet recorded above the fields in the
southern part of the Netherlands
• No earthquakes yet recorded above the fields in Friesland
• We do have some deterministic knowledge on the
mechanism of fault reactivation and subsequent
earthquakes in producing fields (o.a. Phd research Frans
Mulders)
• The ‘winningsplannen’ provide structured information on all
producing fields
Next step is to look for correlation between
field parameters and the recorded tremors
• Look for parameters that have a
good distinguishing capacity.
• Look for threshold values.
• Give a sound physical explanation
why there is a correlation
• Express seismicity in cumulative
released seismic energy per field
log E s  3.81  1.64M L
Used data
Filter dataset by DPfq. Young fields will be
excluded
Hydrocarbon field
Date first quake
DPfq [bar]
Magnitude first quake [-]
Roswinkel
June-92
212
2.7
Bergermeer
August-94
168
3.0
Groningen
December-91
197
2.4
Eleveld
December -86
217
2.8
Bergen
October-01
188
2.7
Annerveen
August -94
266
2.3
Appelscha
June -02
173
1.8
Emmen
October-91
211
2.2
Dalen
August-96
287
1.6
Roden
October -95
203
1.3
VriesNoord
December -96
112
1.9
Ureterp
April-99
171
1.0
Emmen-Nw.A'Dam
September-94
122
1.7
Schoonebeek
December -02
168
1.4
VriesCentraal
July-00
226
1.0
Coevorden
February-97
253
1.2
Filter dataset by DPfq
• Threshold of 72 bar value is calculated
• This value match perfectly the value of 70
bar found by Heriot Watt University
Pers. Com. Kes Heffer Heriot Watt University
Example: no correlation
10000.0000
1000.0000
Released seismic energy [MJoule]
100.0000
10.0000
1.0000
0.1000
0.0100
0.0010
0.0001
0.01
0.1
Porosity [-]
1
Parameter 1: Fault density
fault area1,5 / gross rock volume [-]
• Hypothesis: Induced earthquakes are
generated on weak planes (for
example faults). The probability of
having an earthquake becomes larger
having more faults in the reservoir.
Parameter 1: Fault density
fault area1,5 / gross rock volume [-]
l h
l  h
fault area


gross rock volume
A h
A
1, 5
h:
lb:
A:
1, 5
b
1, 5
1, 5
b
average thickness of the reservoir
total fault length of top reservoir penetrating faults
and boundary faults in a specific reservoir
total area of the top of structure map
Parameter 1: Fault density
reservoir: Sleen
thickness: 50 m
fault length: 4,6 km
area: 5 km2
Result for parameter 1
0,4
reservoir: Coevorden DC
thickness: 80 m
fault length: 129 km
area: 50 km2
Result for parameter 1
8.3
Parameter 1: Fault density
( B3/2 / V) yield = 0.98 ± 0.16
Laplace, Bayes
Parameter 2 Ratio Eburden/Ereservoir
Hypothesis: A relative stiff seal rock compared
to the reservoir rock enhances reactivation of
faults during production.
16
14
normal
RSD [cm]
12
10
8
6
4
2
reverse
0
-2
0
5
10 15 20 25 30 35 40 45
sur
E
Mulders 2003
[GPa]
Parameter 2 Ratio Eburden/Ereservoir [-]
Red: high stiffness contrast
Blue: low stiffness contrast
Parameter 2 Ratio Eburden/Ereservoir
Data
• No static data (lab tests) available out
of the production plans from the
companies. One report from BP on
Platten and Rotliegend in the Bergen
concession
• Inventory of Vp/Vs sonic velocities to
determine dynamic elastic parameters
• Convert to static elastic data
Parameter 2 Ratio Eburden/Ereservoir [-]
(0.93)
(1.34)
Calculate the probability
Van Eijs et al. 2006
Result