Seismic Hazard and Risk Assessment in Groningen - KNGMG
Symposium Induced Seismicity
Seismic Hazard and Risk Assessment in Groningen Symposium on seismicity induced by gas production from the Groningen Field
Jan van Elk & Dirk Doornhof - 1st February 2018
BRON VAN ONZE ENERGIE
Symposium Induced Seismicity
Earthquake studies cover 7 themes
GASWINNING
1
INKLINKEN GASLAAG
2
AARDBEVINGEN
3
GRONDBEWEGING
HAZARD
4
BLOOTSTELLLING HUIZEN EN MENSEN
5
STERKTE VAN HUIZEN
6
VEILIGHEID
7
Building Damage
RISK
Symposium Induced Seismicity
Field Measurements and Monitoring
Subsidence Satellite & Network of Reflector-GPS Stations Gas production and pressure
Flexible Geophone Network Ground Movement
Moveable Geophone Spread
70 Accelerometers (KNMI)
Compaction and Faults Rock Core and In-situ Compaction Monitoring
Earthquakes 70 Shallow Geophone Strings (KNMI)
Subsidence 10 GPS stations
Building Damage Trend Analysis
Earthquakes
Building Vibrations
Gravity
Deep Geophone Strings
300+ Building Sensors (TNO)
92 locations
Symposium Induced Seismicity
Introduction Hazard and Risk Assessment The hazard- and risk assessment spans from cause (gas production) to effect (accidents, harm and building damage).
The uncertainties in each step of the assessment are identified, estimated and consistently incorporated in the assessment.
A traditional Probabilistic Seismic Hazard and Risk Framework is used (based on Cornell, 1968).
Implementation is based on Monte Carlo Method (C- and Python Code) NAM has sought the assistance and advice of external experts from academia and knowledge institutes for each expertise area. Rigorous assurance processes are in
place.
Key is the collection of data in Groningen to prepare a hazard and risk assessment specific to the Groningen situation.
4
Symposium Induced Seismicity
Gas Production Detailed mapping of faults in the reservoir. This forms the basis of geomechanical studies into fault behaviour (e.g. University Utrecht).
Reservoir Model has been history matched using downhole pressure, converted closed-in THP, waterencroachment (PNL) and subsidence. Evaluated model Slope (scales clipped)
performance against gravity survey data.
Optimisation of the distribution of the gas production from the field to reduce seismicity.
Symposium Induced Seismicity
Seismogenic Model G Topographic
Pore DP pressure depletion
Learn
Forecast
gradient
ezz Vertical
Ci Initial
strain
Coulomb stress
Physics-based seismogenic models of increasing complexity have been evaluated using prospective testing.
Theory of extreme threshold failures within a heterogeneous poro-elastic thin-sheet forecasts Groningen induced seismicity.
Exponential shear strain trend with ETAS aftershocks.
Symposium Induced Seismicity
Ground Motion Model to predict distributions—medians plus sigmas—of Sa(T), PGV and duration (DS5-75) as needed for risk assessments.
Applicable from ML 2.5 to largest Mmax, accounting for finite rupture dimensions of larger events and epistemic uncertainty associated with extrapolation from smallmagnitude recordings.
Model the variation of near-surface profiles across the field and the non-linear response of soft soil deposits.
Model to reflect the unique velocity structure above the gas reservoir. Model to reflect source characteristics of Groningen earthquakes—and potential for larger stress drops for bigger event.
Symposium Induced Seismicity
Ground Motion
Symposium Induced Seismicity
Seismic Hazard Curves
Loppersum
Groningen
Mean hazard curve 95% prediction interval
PGA [g]
PGA [g]
Symposium Induced Seismicity
Seismic Hazard Maps Assessment period: 1-1-2017 to 1-1-2022 Production scenario: 24 bcm/year Exceedance probability: 0.21%/year (Poisson return periods 475 year)
PGA
PGV
Max = 0.166 g
Max = 0.122 m/s
Symposium Induced Seismicity
Exposure in Groningen
Gebouw dichtheid Building Density
Population Density - Daytime
Figure 10: Population densityofexpressed as a day-night average and measured FigureTHE 4: The number of buildings per square kilomet re as measured on a GEM (Global Earthquake Model) Taxonomy Structural Systems is used to −2 a 250 bydensity 250 m isregular 250 by 250 m regular grid. T he maximumonnumber 12300grid. km This is based on a population data for individual building recorded foundclassify wit hin the citybuildings of Groningen (G). Teach he let ters ’D’,building ’E’, ’H’,as’L’, ’W’ in the exposure dat abase (Free et al., the in Groningen into typologies. 2013a,b). Grey denote buildings currently without these population data. denote the place names Delfzijl, Eemshaven, Hoogezand, Loppersum and region es buildings currently without a classificat ion Winschoten respectively, and t he black lineTheremaining denotes t he grey out line of denot t he field. Map coordinat es are labelled in kilomet res.of building class. The letters ’D’, ’E’, ’H’, ’G’, ’L’, ’W’ denote the place names
Symposium Induced Seismicity
Building Response to Earthquakes In-situ material characterisati on
13 URM houses 2 RC buildings
Lab material characterisation
≈ 200 test specimens (taken from actual houses)
Components testing
7 URM walls in-plane 8 RC precast connections (2-way) 3 URM walls OOP one-way 5 URM walls OOP two-way (damage)
Full-structure testing
2 URM houses (shake-table) 1 URM houses (damage, collapse) 2 URM structures (push-over) 1 roof + gables (damage, collapse) 1 roof (cyclic, collapse) 2 RC structures (cyclic, damage) 1 RC structures (shake-table)
Seismic building response study program consists of:
In-situ testing Building material testing in laboratory Testing of small assemblages Testing of walls Testing of full Building Structures
Eucentre (Italy) and LNEC (Portugal) ARUP TU Delft and TU Eindhoven MOSAYK
Partners in the program are:
Experiments are designed to improve and
calibrate the modelling of Building Response Rigorous pre- and post-diction approach
Symposium Induced Seismicity
Building Response to Earthquakes LNEC, Portugal
Eucentre, Italy
Floor Accelerogram input at
LNEC
Symposium Induced Seismicity
Building Response to Earthquakes East Wall
RNDM EQ1 RNDM EQ1 RNDM EQ1 EQ1 RNDM EQ1 RNDM EQ2 EQ2 EQ2 EQ2 RNDM EQ2 RNDM EQ2 EQ2 RNDM EQ2 RNDM EQ2 RNDM
RNDM-01 EQ1-25% RNDM-03 EQ1-50% RNDM-05 EQ1-50%-C EQ1-100% RNDM-08 EQ1-150% RNDM-10 EQ2-30%-C EQ2-30%-C EQ2-30%-C EQ2-50% RNDM-15 EQ2-100% RNDM_17 EQ2-50%-C EQ2-125% RNDM-20 EQ2-150% RNDM-22 EQ2-200% RNDM-24
0.050 0.024 0.050 0.049 0.050 0.048 0.096 0.050 0.144 0.050 0.053 0.053 0.053 0.079 0.050 0.159 0.050 0.079 0.199 0.050 0.239 0.050 0.319 0.050
0.024 0.050 0.050 0.099 0.137 0.064 0.059 0.056 0.087 0.170 0.114 0.194 0.243 0.307 -
0.015 0.031 0.031 0.056 0.077 0.049 0.045 0.043 0.067 0.123 0.088 0.133 0.164 0.218 -
EQ2-100%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
North Wall
South Wall
Nominal Recorded Calculated Nominal Calculated Calculated Calculated PGA PGA PGV Sa(T1) Sa(T1) Sd(T1) mHI [g] [g] [m/s] [g] [g] [mm] [mm]
0.049 0.097 0.089 0.179 0.268 0.081 0.081 0.081 0.122 0.245 0.122 0.306 0.367 0.489 EQ2-125%
Test Name
EQ2-150%
Test Input
EQ2-200%
Test #
West Wall
0.055 0.126 0.108 0.229 0.369 0.096 0.087 0.083 0.125 0.286 0.183 0.324 0.404 0.654 -
0.4 0.9 0.8 1.6 2.6 0.7 0.6 0.6 0.9 2.1 1.3 2.3 2.9 4.7 -
8.3 18.2 17.3 34.9 47.7 23.4 22.2 21.5 31.8 62.1 41.3 69.0 84.4 111.6 -
14
Symposium Induced Seismicity
Building Response to Earthquakes Masonry
Concrete
Symposium Induced Seismicity
Building Response to Earthquakes Modelling pre-and post-diction done by: ARUP using LS-Dyna MOSAYK using ELS - Extreme Loading for
Structures TU Delft using Diana EUCentre using Tremuri
Symposium Induced Seismicity
Seismic Risk Risk Assessment allows comparison with the Meijdam-Norm for Local Personal Risk (LPR).
No buildings are exposed to mean LPR > 10-4.
Some 2,800 houses have 10-5<mean LPR
Seismic Hazard and Risk Assessment in Groningen Symposium on seismicity induced by gas production from the Groningen Field
Jan van Elk & Dirk Doornhof - 1st February 2018
BRON VAN ONZE ENERGIE
Symposium Induced Seismicity
Earthquake studies cover 7 themes
GASWINNING
1
INKLINKEN GASLAAG
2
AARDBEVINGEN
3
GRONDBEWEGING
HAZARD
4
BLOOTSTELLLING HUIZEN EN MENSEN
5
STERKTE VAN HUIZEN
6
VEILIGHEID
7
Building Damage
RISK
Symposium Induced Seismicity
Field Measurements and Monitoring
Subsidence Satellite & Network of Reflector-GPS Stations Gas production and pressure
Flexible Geophone Network Ground Movement
Moveable Geophone Spread
70 Accelerometers (KNMI)
Compaction and Faults Rock Core and In-situ Compaction Monitoring
Earthquakes 70 Shallow Geophone Strings (KNMI)
Subsidence 10 GPS stations
Building Damage Trend Analysis
Earthquakes
Building Vibrations
Gravity
Deep Geophone Strings
300+ Building Sensors (TNO)
92 locations
Symposium Induced Seismicity
Introduction Hazard and Risk Assessment The hazard- and risk assessment spans from cause (gas production) to effect (accidents, harm and building damage).
The uncertainties in each step of the assessment are identified, estimated and consistently incorporated in the assessment.
A traditional Probabilistic Seismic Hazard and Risk Framework is used (based on Cornell, 1968).
Implementation is based on Monte Carlo Method (C- and Python Code) NAM has sought the assistance and advice of external experts from academia and knowledge institutes for each expertise area. Rigorous assurance processes are in
place.
Key is the collection of data in Groningen to prepare a hazard and risk assessment specific to the Groningen situation.
4
Symposium Induced Seismicity
Gas Production Detailed mapping of faults in the reservoir. This forms the basis of geomechanical studies into fault behaviour (e.g. University Utrecht).
Reservoir Model has been history matched using downhole pressure, converted closed-in THP, waterencroachment (PNL) and subsidence. Evaluated model Slope (scales clipped)
performance against gravity survey data.
Optimisation of the distribution of the gas production from the field to reduce seismicity.
Symposium Induced Seismicity
Seismogenic Model G Topographic
Pore DP pressure depletion
Learn
Forecast
gradient
ezz Vertical
Ci Initial
strain
Coulomb stress
Physics-based seismogenic models of increasing complexity have been evaluated using prospective testing.
Theory of extreme threshold failures within a heterogeneous poro-elastic thin-sheet forecasts Groningen induced seismicity.
Exponential shear strain trend with ETAS aftershocks.
Symposium Induced Seismicity
Ground Motion Model to predict distributions—medians plus sigmas—of Sa(T), PGV and duration (DS5-75) as needed for risk assessments.
Applicable from ML 2.5 to largest Mmax, accounting for finite rupture dimensions of larger events and epistemic uncertainty associated with extrapolation from smallmagnitude recordings.
Model the variation of near-surface profiles across the field and the non-linear response of soft soil deposits.
Model to reflect the unique velocity structure above the gas reservoir. Model to reflect source characteristics of Groningen earthquakes—and potential for larger stress drops for bigger event.
Symposium Induced Seismicity
Ground Motion
Symposium Induced Seismicity
Seismic Hazard Curves
Loppersum
Groningen
Mean hazard curve 95% prediction interval
PGA [g]
PGA [g]
Symposium Induced Seismicity
Seismic Hazard Maps Assessment period: 1-1-2017 to 1-1-2022 Production scenario: 24 bcm/year Exceedance probability: 0.21%/year (Poisson return periods 475 year)
PGA
PGV
Max = 0.166 g
Max = 0.122 m/s
Symposium Induced Seismicity
Exposure in Groningen
Gebouw dichtheid Building Density
Population Density - Daytime
Figure 10: Population densityofexpressed as a day-night average and measured FigureTHE 4: The number of buildings per square kilomet re as measured on a GEM (Global Earthquake Model) Taxonomy Structural Systems is used to −2 a 250 bydensity 250 m isregular 250 by 250 m regular grid. T he maximumonnumber 12300grid. km This is based on a population data for individual building recorded foundclassify wit hin the citybuildings of Groningen (G). Teach he let ters ’D’,building ’E’, ’H’,as’L’, ’W’ in the exposure dat abase (Free et al., the in Groningen into typologies. 2013a,b). Grey denote buildings currently without these population data. denote the place names Delfzijl, Eemshaven, Hoogezand, Loppersum and region es buildings currently without a classificat ion Winschoten respectively, and t he black lineTheremaining denotes t he grey out line of denot t he field. Map coordinat es are labelled in kilomet res.of building class. The letters ’D’, ’E’, ’H’, ’G’, ’L’, ’W’ denote the place names
Symposium Induced Seismicity
Building Response to Earthquakes In-situ material characterisati on
13 URM houses 2 RC buildings
Lab material characterisation
≈ 200 test specimens (taken from actual houses)
Components testing
7 URM walls in-plane 8 RC precast connections (2-way) 3 URM walls OOP one-way 5 URM walls OOP two-way (damage)
Full-structure testing
2 URM houses (shake-table) 1 URM houses (damage, collapse) 2 URM structures (push-over) 1 roof + gables (damage, collapse) 1 roof (cyclic, collapse) 2 RC structures (cyclic, damage) 1 RC structures (shake-table)
Seismic building response study program consists of:
In-situ testing Building material testing in laboratory Testing of small assemblages Testing of walls Testing of full Building Structures
Eucentre (Italy) and LNEC (Portugal) ARUP TU Delft and TU Eindhoven MOSAYK
Partners in the program are:
Experiments are designed to improve and
calibrate the modelling of Building Response Rigorous pre- and post-diction approach
Symposium Induced Seismicity
Building Response to Earthquakes LNEC, Portugal
Eucentre, Italy
Floor Accelerogram input at
LNEC
Symposium Induced Seismicity
Building Response to Earthquakes East Wall
RNDM EQ1 RNDM EQ1 RNDM EQ1 EQ1 RNDM EQ1 RNDM EQ2 EQ2 EQ2 EQ2 RNDM EQ2 RNDM EQ2 EQ2 RNDM EQ2 RNDM EQ2 RNDM
RNDM-01 EQ1-25% RNDM-03 EQ1-50% RNDM-05 EQ1-50%-C EQ1-100% RNDM-08 EQ1-150% RNDM-10 EQ2-30%-C EQ2-30%-C EQ2-30%-C EQ2-50% RNDM-15 EQ2-100% RNDM_17 EQ2-50%-C EQ2-125% RNDM-20 EQ2-150% RNDM-22 EQ2-200% RNDM-24
0.050 0.024 0.050 0.049 0.050 0.048 0.096 0.050 0.144 0.050 0.053 0.053 0.053 0.079 0.050 0.159 0.050 0.079 0.199 0.050 0.239 0.050 0.319 0.050
0.024 0.050 0.050 0.099 0.137 0.064 0.059 0.056 0.087 0.170 0.114 0.194 0.243 0.307 -
0.015 0.031 0.031 0.056 0.077 0.049 0.045 0.043 0.067 0.123 0.088 0.133 0.164 0.218 -
EQ2-100%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
North Wall
South Wall
Nominal Recorded Calculated Nominal Calculated Calculated Calculated PGA PGA PGV Sa(T1) Sa(T1) Sd(T1) mHI [g] [g] [m/s] [g] [g] [mm] [mm]
0.049 0.097 0.089 0.179 0.268 0.081 0.081 0.081 0.122 0.245 0.122 0.306 0.367 0.489 EQ2-125%
Test Name
EQ2-150%
Test Input
EQ2-200%
Test #
West Wall
0.055 0.126 0.108 0.229 0.369 0.096 0.087 0.083 0.125 0.286 0.183 0.324 0.404 0.654 -
0.4 0.9 0.8 1.6 2.6 0.7 0.6 0.6 0.9 2.1 1.3 2.3 2.9 4.7 -
8.3 18.2 17.3 34.9 47.7 23.4 22.2 21.5 31.8 62.1 41.3 69.0 84.4 111.6 -
14
Symposium Induced Seismicity
Building Response to Earthquakes Masonry
Concrete
Symposium Induced Seismicity
Building Response to Earthquakes Modelling pre-and post-diction done by: ARUP using LS-Dyna MOSAYK using ELS - Extreme Loading for
Structures TU Delft using Diana EUCentre using Tremuri
Symposium Induced Seismicity
Seismic Risk Risk Assessment allows comparison with the Meijdam-Norm for Local Personal Risk (LPR).
No buildings are exposed to mean LPR > 10-4.
Some 2,800 houses have 10-5<mean LPR
