Introduction &Recent Research in Groningen Earthquakes
İhsan Engin BAL
Research Group on Earthquake Resistant Structure
Displacement-Based Earthquake Loss Assessment: Method Development and Application to Turkish Building StockI.E. Bal, H. Crowley, R. PinhoResearch Report Rose 2010/02
Motivation for earthquake engineering…
THE RESPONSE OF UNREINFORCED MASONRY
TO LOW-AMPLITUDE RECURCIVE LOADS:
CASE OF GRONINGEN GAS FIELD
COMPDYN 2017
İhsan E. Bal, Eleni Smyrou and Dimitrios Dais
Introduction to Induced Seismicity
• The largest gas field in Europe
and 10th in the world
• No earthquakes were reported
from the Groningen area prior to
1991
• Before smaller than the humanly
perceptible levels
In 2014 Oklahoma actually surpassed the San Francisco region, famous for its seismic excitations, in terms of the rate of earthquakes
Oklahoma region is lacking of any important fault
Introduction to Induced Seismicity – Case of Oklahoma
low magnitude normally cannot be perceived by humans
Induced seismicity events are felt due to their shallow depth
(e.g. for Groningen the depth is 3 km)
Induced vs Deep Natural Earthquakes
• The gas field was discovered in 1959
• Gas Initially In Place (GIIP): 3000 billion m3 (bcm)
• Until 2015, 75% of the GIIP had been extracted
• Annual number of earthquakes has been
increasing since 2003
• Correlation between gas production - seismicity
Groningen Gas Field
Low magnitude Excitation
vs
Moderate Scale Damages
Motivation of the Study– Groningen Paradox
Damages in the Region
• 60,000 homes lie within the earthquake
zone
• 40,000 properties are considered at risk of
damage and are under inspection
• 6,000 damage claims
• Value of properties has decreased
drastically
• Rising uncertainty
• 1-story or 2-story houses compose the main building stock, mostly URM
• Lack of seismic design – no seismic loading was expected
• Low normal stresses (low shear strength) on the load bearing and veneer walls
• Limited experimental research on the low-cycle fatigue of masonry under recursive loads
• Further investigation of URM response in low-amplitude range is required
• Shallow quakes (3km depth) release more energy at the surface - limited area that is
affected, but higher intensity is attained
Particularities of the Groningen Problem
• Low-rise and stiff structures with low natural period
• Resonance between the excitation and the structures (?)
• Poor soil conditions (soft clay with 40kPa allowable strength !)
• Unlikely with most of the similar cases, Groningen gas field is densely populated
Motivation of the Study– Groningen Paradox
Maximum intensity : VI
magnitude Mw : 3.6
Recorded PGA : 85 cm/s2
From Codes : 10 cm/s2
Recorded PGV : 3.45 cm/s
Focal Depth: 3 km (quite shallow)(EMS98 intensity grades)
2012 Huizinge (Groningen) EQ, Mw=3.6
The August 16th, 2012, Earthquake Near Huizinge (Groningen)
PGA 85 cm/s2
T : 0 - 0.5 sec
Wavelet Analysis (Tectonic Event on Soft Soil)
Smyrou E., Bal İ. E., Tasiopoulou P., and Gazetas G., “Wavelet analysis for relating soil amplification and liquefaction effects with seismic performance of precast structures, ”, Earthquake Engineering and Structural Dynamics, (2016)
Wavelet analysis from a tectonic earthquake recorded on soft soil
Frequency Issue
Wavelet Analysis (Induced Seismicity)
2012 EQ, Groningen 2014 EQ, Groningen
Wavelet Analysis (Tectonic Events)
2012 EQs, Emilia-Romagna, Italy
The August 16th, 2012, Earthquake Near Huizinge (Groningen)
Before the Event
According to studies, induced seismic events with M < 3.9
NO significant structural damage
Low Risk
The August 16th, 2012, Earthquake Near Huizinge (Groningen)
After the Event
safety of the citizens might be at risk
Updated seismic hazard analyses : Mmax = 5
Quite a Catastrophic Scenario
Available Experiments on Groningen-type URM
Sponsored by NAM
Conducted by EUCENTRE
(Pavia, Italy)
In-Plane Cyclic Shear-
Compression Tests
Slender Squat
Available Experiments on Groningen-type URM
Fy = 60 kN, δy = 0.45 mm, δu = 8.2 mm
Response of Squat Wall
Available Experiments on Groningen-type URM
max SA (Huizinge, 2012) = 2.2 m/sec2
mass = 13 tn
Squat Wall
Fb = m * SA = 28.6 kN << Fy = 60 kN
ZERO damage
would be expected
BUT…
Available Experiments on Groningen-type URM
Full-scale Building Test First Cracks
test #16 : PGA = 1.6 m/s2
PGA (Huizinge, 2012) = 0.85 m/sec2 << test #16 : PGA = 1.6 m/s2
Available Experiments on Groningen-type URM
Response of Squat Wall
1st cycle
Stress Drop in Small Cycles
Cyclic response in large cycles
Force
Displacement
Force
Displacement
Cyclic response in small cycles
Available Experiments on Groningen-type URM
first 3x3 cyclesidealized hysteresis
backbone (still working on it !)
1) Δfi
2) αΚi
3) dres
Exercise with Existing Hysteresis Loops – Cyclic Loading
Ramberg–Osgood Model in
SeismoStruct Software
Exercise with Existing Hysteresis Loops – Cyclic Loading
Next Steps
I. E. Bal et al., 2010
Exercise with Existing Hysteresis Loops – Seismic Loading
Exercise with Existing Hysteresis Loops – Seismic Loading
• Full database for 2014
• Proximity to Huizinge epicenter
• Distribution of earthquakes
(1 M3 & 18 1.5≤M<3)
Exercise with Existing Hysteresis Loops – Seismic Loading
Seismic Scenarios
1) Huizinge (2012)
2) 1-year + Huizinge (2012)
3) 5-year + Huizinge (2012)
4) 10-year + Huizinge (2012)
Exercise with Existing Hysteresis Loops – Seismic Loading
Analysis
δmax : 0.06 mm
Test
δy = 0.45 mm & δu = 8.2 mm
NOT destructive acting on its own
Exercise with Existing Hysteresis Loops – Seismic Loading
Conclusions
Conclusions
• Experimental evidence show that the response of the Groningen-type URM walls is quite
different in the very first cycles then in the post-yield phase
• The available hysteresis loops are designed to capture more accurately the post-yield phase,
and thus are not suitable for the small amplitude cycles
• The most well calibrated model was not able to present any cumulative damage, and even
has tendency to re-center, contradicting the expectations and previous experimental
findings under normal stress cases
Exercise with Existing Hysteresis Loops – Seismic Loading
• There is a characteristic force drop in the small amplitude cycles in case of unloading, which
cannot be modeled by using the widely available common hysteresis loops
• A special hysteresis rule that represents not only the large displacement and the post-yield
phase but also the very small amplitudes need to be developed.
Thank you