Content uploaded by Dimitris Dais
All content in this area was uploaded by Dimitris Dais on Sep 14, 2018
Content may be subject to copyright.
Monitoring, Assessment and Diagnosis
of Fraeylemaborg in Groningen, Netherlands
, Eleni Smyrou
, Ihsan Engin Bal
and Jelle Pama
Newcastle University, Newcastle upon Tyne, UK
Hanze University of Applied Sciences Groningen, Groningen, Netherlands
Structural Design, Hanze University of Applied Sciences Groningen,
Abstract. Fraeylemaborg is a noble house in an earthquake-stricken area of the
Netherlands due to the induced seismicity events in the region. The structure is
located in the middle of the town of Slochteren which gave its name to the
largest gas ﬁeld in the world upon its discovery in 1959. The gas extraction has
caused small-magnitude shallow earthquakes during the last decade, damaging
not only the residential inventory but also the historical structures in the area.
The main building of Fraeylemaborg sits on an artiﬁcial island surrounded by
water channels, rendering the problem of earthquake response even more
complicated. A small part of the main structure on the island was built in the
century, while the construction of additional parts and morphological
alterations had taken place until the 18
century. The structure has been sub-
jected to several small magnitude earthquakes causing damages on the load
bearing system. An extensive renovation and repair of damages took place in
recent years, however the latest seismic events imposed again damage to the
structure. This paper presents a project of monitoring, assessment and diagnosis
of problems for the Fraeylemaborg, the most important “borg”of the region,
underlining the particularities of the induced seismicity problem. The FE model
has been calibrated by using ambient vibration tests. Combination of earthquake
and soil settlement loads have been applied on the calibrated model. The paper
develops scenarios that help in explaining the reasons behind the damages on
this structure during the recent shallow and low-magnitude induced seismicity
Keywords: Seismic response Induced seismicity Structural monitoring
Ambient vibration tests
Groningen is the largest gas ﬁeld in Europe and 10th in the world. Due to the extensive
gas extraction induced earthquakes of relatively “larger”magnitude have been recorded
in the very last decade. The building stock in the region comprises single- and two-
storey unreinforced masonry (URM) houses constructed with no seismic considerations.
R. Aguilar et al. (Eds.): Structural Analysis of Historical Constructions,
RILEM Bookseries 18, pp. 2188–2196, 2019.
URM responds to cyclic load reversals in a non-ductile way unless special mea-
sures are taken. Moreover, low normal stresses on the load bearing and veneer walls in
conjunction with the recursive nature of the seismic actions, irrespective of their low
amplitude, can lead to micro damage .
The herein study focuses on the assessment of the current condition of Fraeyle-
maborg, an emblematic historical building of the region that has suffered damage due
to several earthquakes, among which the Slochteren event (2017) that its epicenter lied
only 2 km away from Fraeylemaborg.
2 Fraeylemaborg and Its Damage History
The historical building of Fraeylemaborg in Slochteren sits on the gas ﬁeld of the
Groningen region. Fraeylemaborg (Fig. 1) was built in 1300 as an austere stone house.
In the 16
century, the house expanded and was embellished, and a moat around the
structure was dug. After 1670 the two wings were added giving its U-shaped shape,
while in 1781, a major renovation took place that determined the current appearance of
the mansion . The structure is composed of thick clay brick walls and wooden ﬂoors.
The plan of the building is 28.6 m long and 22.6 m wide. The top of the tower reaches
20 m height. The tower did not exist in the initial state of the building but was added in
a later phase, therefore, there is a partial connection of the tower to the building,
namely, only two sides of the tower are connected to the adjacent walls of the structure
while the other two simply sit on the wooden ﬂoor. The extensive damage on the
building (Fig. 2) due to the recent seismic activity (Table 1) has been repaired so no
cracks on the building walls were visible at the time of assessment.
3 Damage Scenario, Field Measurements and Numerical
The cracks on the structure, both horizontal and vertical, were scattered along the walls
without any distinctive pattern, atypical for cracks caused by a seismic excitation which
tend to develop diagonally accumulating at the wall bottom and extending along the wall
height. Moreover, the small cracks on the retaining walls and the open bended shape of
the vertical rod of the wall anchors (Fig. 2d) evidence movement of the retaining wall
followed by the soil beneath and resulting in differential displacements and settlement of
the structure. These observations led to the hypothesis that soil movements initiated by
earthquake events, not the seismic loading itself, inﬂicted damage, correlating thus the
damage patterns with the deformability of the foundation .
Fig. 1. Fraeylemaborg
Monitoring, Assessment and Diagnosis of Fraeylemaborg 2189
For validating the aforementioned hypothesis, a detailed 3-D numerical model
(Fig. 3) was constructed utilizing the Finite Element software Abaqus . Solid 10-
node quadratic tetrahedron elements (C3D10) were used. The structural elements, i.e.
the clay brick walls and wooden ﬂoors, were modeled as solid homogenous members
with isotropic elastic properties.
A rather simple experimental study was carried out to obtain some rough material
properties of the existing structure: Six bricks retrieved during the previous restoration
works were subjected to compression tests. Their compressive strength was 0.25 MPa
in average (standard deviation 0.052 MPa), a value considerably lower than those
obtained from recent experimental studies [5,6] on clay bricks currently used in the
construction in Groningen. Due to lack of any other material tests the elastic material
properties of the URM walls and the wooden ﬂoors were calibrated based on the
ﬁndings from in-situ ambient vibration measurements carried out on the building
within this study. Force-balanced accelerometer sensors with ultra-low self-noise, both
in uniaxial and triaxial versions, were placed at the top ﬂoor (Fig. 4) and the top of the
tower in appropriate locations, while the motion at the foundation level was assumed
zero. The sensors were ﬁxed on the brick walls as the wooden ﬂoor did not provide a
Fig. 2. (a) View from North towards South (in red the wing with extensive damage), (b & c)
Wall cracks, (d) Cracks on the retaining wall and detachment of the vertical rod of the anchors
that constrain the wall to the soil layer beneath.
Table 1. The main events recorded in the Groningen region after the Huizinge event (2012).
Place Date Lat [°] Lon [°] Depth [km] M PGA [cm/s
Huizinge 16-08-2012 53.345 6.671 3 3.6 (M
Zeerijp 08-01-2018 53.363 6.751 3 3.4 (M
‘t Zandt 13-02-2014 53.357 6.782 3 3.0 (M
Zandeweer 05-11-2014 53.374 6.678 3 2.9 (M
Slochteren 27-05-2017 53.211 6.834 3 2.6 (M
2190 D. Dais et al.
The measurements from the ambient vibration analysis were processed with
Artemis software . Operational Modal Analysis (OMA) was carried out by means of
Frequency Domain Decomposition (FDD) technique enabling the acquisition of the
natural frequencies (Fig. 5) and the corresponding mode shapes, directly from the raw
measured time series data of the structure under natural conditions.
The numerical model was then calibrated to provide similar ﬁrst modes and mode
shapes, achieving a good match as shown in Fig. 5. The modulus of elasticity of the
URM walls was derived as E = 1 GPa. The difference between this value and the value
of 4.7 GPa estimated after tests on Groningen resembling wall specimens [5,6]is
justiﬁable considering that materials used some centuries ago are expected to have
considerably lower properties than similar modern materials.
The subsequent modes represent local modes. The wooden ﬂoors do not behave as
rigid diaphragms, thus a semi-ﬂexible assumption simulates more accurately the real
case. The displacement values corresponding to the ﬁrst mode deformed shape, as
computed by both Abaqus and Artemis analyses, are presented in Fig. 6exhibiting a
satisfactory match between ﬁeld and numerical analysis results. The values indicated
have been normalized with respect to the maximum value of deformation obtained
from the sensors placed at the top ﬂoor. The primary translational direction is along y
axis. The structure in the ﬁrst mode shape oscillates in the y direction and the x
component is negligible, thus the x component is not presented in the ﬁgure.
Fig. 3. The 3-D numerical model in Abaqus (the clay brick walls in maroon color, wooden ﬂoor
Fig. 4. Locations of the accelerometer sensors during ambient vibration measurements. The
blue arrows indicate the direction the sensors recorded.
Monitoring, Assessment and Diagnosis of Fraeylemaborg 2191
The OMA results highlight also that the tower does not follow the main structure in
y direction, while in x direction its displacements are not negligible. Such a response is
explicable considering the connection of the tower to the main walls beneath, as
already mentioned, nevertheless, it is difﬁcult to be captured by the numerical model in
which the tower was modelled as fully ﬁxed to rest of the building. At this point the
response of the tower attached on the top ﬂoor of the structure is explained. According
to the results from the Artemis analysis, the tower in the ﬁrst mode seems to deform
incoherently with the rest of the building, while the component of displacement is not
negligible in the x direction. This behavior can be explained by the way that the tower
is connected to the building. For the scope of this study, the ﬁxed connection is deemed
adequate, however, the real response of the tower should be taken into account in the
ﬁnal assessment of the response of the structure.
Fig. 5. The ﬁrst six modes of the model as obtained with Artemis software by means of
Frequency Domain Decomposition (FDD) (a) and comparison of the ﬁrst 6 modes as calculated
by Abaqus and Artemis Modal software (b).
Fig. 6. Normalised displacements in y direction for the 1
mode shape derived from Artemis
(red) and Abaqus (black). Note: Sensors at positions 3 and 5 are uniaxial and record only in x
2192 D. Dais et al.
4 Numerical Analyses
Since the validated numerical model provides a realistic estimation of the response of
the structure, analyses were conducted to gain further insight. In particular, the fol-
lowing scenarios were considered:
•Settlement in the North side of the structure
•Static equivalent horizontal earthquake load
•Settlement in the North side of the structure and then static equivalent horizontal
As there are no measurements of the soil deformations, a possible settlement pattern
was regarded. As explained above, a likely opening of the front (North side) retaining
walls would impose a possible lateral and/or vertical movement of the foundations.
This may be a plausible explanation for the extensive cracks in the front (North) walls
of the structure, after seismic events. A soil movement of such would result in dif-
ferential settlements in the front part of the structure.
The seismic hazard study for the exact location of the structure were taken into
consideration in order to have an estimation of the seismic forces. A very thorough grid
of site-speciﬁc parameters has been established for the calculation of the elastic design
response spectrum  in correspondence with the newly developed Dutch seismic
guidelines (Draft NPR 9998:2017) . The obtained response spectra for return periods
475 and 2475 years are presented in the Fig. 7. Speciﬁcally, for each return period, two
response spectra are created: one coming from the exact site-speciﬁc response (UHS
data) and one from the smooth design spectrum as calculated by the Draft NPR
9998:2017 parameters and speciﬁcation. The ﬁrst mode shape of the structure (T1 = 0.
35 s as calculated by the analysis in Abaqus) falls into the plateau of the design spectra
according to the existing seismic codes and the spectral acceleration is 0.62 g. This
acceleration was imposed to the building in a quasi-static way. It should be noted that
the main purpose of this paper is to explain the possible causes of the observed
damages, but not to ﬁnd the accurate damage distribution in the building for a speciﬁc
seismic load level. In this respect, the seismic acceleration levels applied to the
structure during the numerical analyses are rather indicative.
In this paper only the results for acceleration in the x direction of the building will
be shown. In Fig. 7, the response spectra from the strongest record from the earthquake
events Huizinge (2012), Zeerijp (2018) and Slochteren (2017) are highlighted as well.
The waveforms were retrieved from the Seismic Network dataset of KNMI . The
epicenter of the Slochteren (2017) record (PGA 34 cm/sec
) was about 2 km from
Fraeyelemaborg and it caused some limited damages to the walls of the structure. It
ought to be noted that for the records from the 3 seismic events the direction of the
maximum PGA was calculated and the corresponding response spectra are presented in
Fig. 7. More speciﬁcations about the calculation of the direction of the maximum PGA
can be found in the study by Smyrou et al. 
The third and last scenario that was taken into consideration was the combination of
the ﬁrst and the second scenario. Precisely, ﬁrstly the settlement pattern was imposed
and then the seismic loading was taken into effect.
Monitoring, Assessment and Diagnosis of Fraeylemaborg 2193
In Fig. 8, the response of the North-West wall of the structure is presented for the
scenario (1). The location of the aforementioned wall can be seen in the Fig. 8(a) and
(b). The contour of the maximum principle stresses shows that the regions with the
maximum stresses are in good agreement with the damage pattern traced in the real
structure, as shown in Fig. 8(a) and (b).
Fig. 7. The response spectra as prescribed by the new-developed seismic code of the
Netherlands (NPR)  and from the exact site-speciﬁc response (UHS data). The response
spectra from the strongest record from the earthquake events Huizinge (2012), Zeerijp (2018) and
Slochteren (2017) are highlighted as well.
Fig. 8. (a) The crack pattern observed in North-West “wing”of the structure, (b) the response of
the same wall for the scenario (1), (c) the response of the structure for the scenario (2) (North
side) and (d) the response of the structure for the scenario (3) (North side). For (b), (c) and (d) the
contours show the maximum principal strengths with the grey parts depicting the locations that
the stresses are evolving excessively.
2194 D. Dais et al.
For the scenario (2), the same results are shown in the Fig. 8c. It can be easily
inferred that the regions of the maximum stresses are in the diagonal direction –a
typical crack pattern for URM under seismic excitation. As it was noted above, this is
the expected crack pattern if the damage was caused solely by the earthquake load.
Lastly, for the scenario (3), it can be stated that the regions with the possible
damage extend in a wider part of the front walls as in comparison with the scenario (2).
Thus, it is evident that the imposed settlements have impact on the response of the
structure under seismic loading. More speciﬁcally, it must be taken into consideration
that for the design earthquake load the potential damage will be more extensive for the
case that the building has already suffered some settlements.
Fraeylemaborg experienced severe damages during the past seismic activities. This
study combines the collected data and the numerical analyses results to ﬁnd out pos-
sible scenarios for explaining the damage.
The numerical models used in the study were calibrated by using the ambient
vibration test results. It was found out that the modulus of elasticity of the brick
masonry is up to 5 times lower than the reported value for the residential URM houses
in the region. This is also in line with the compression tests conducted on representative
bricks from the structure. Furthermore, the numerically determined mode shapes and
vibration periods are in good agreement with the ﬁeld measurements.
The emblematic tower of the structure sits on the bearing walls not with full ﬁxity
but with a relatively loose support system. The tower thus found in the ambient test
results to move separately than the rest of the structure, an important aspect that needs
to be reﬂected in future numerical models.
The structural cracks observed in the past events were compared with numerically
created loading simulations on the calibrated computer model of the structure. It was
found that the lateral and/or vertical soil movement along the URM retaining walls,
triggered by the seismic events, can be a reasonable scenario that occurred during the
latest seismic events.
Acknowledgments. The authors would like to thank the board of Fraeylemaborg Foundation,
and speciﬁcally Mrs. Marjon Edzes and Mr. Gerard de Haan for their help and cooperation.
Fugro ofﬁces in Groningen and in Istanbul are also acknowledged for their insights into the soil
properties in the region. Special thanks to Mrs. Berﬁn Yardak, Mr. Jelmer Bakker and Mr.
Remco van den Belt, students of Hanze University of Applied Sciences, for the valuable data
collected during their undergraduate project. Finally, the ﬁnancial support from EPI Kennis-
centrum in Groningen is gratefully acknowledged.
Monitoring, Assessment and Diagnosis of Fraeylemaborg 2195
1. Dais D, Bal IE, Smyrou E (2017) Discussion on the response of unreinforced masonry to
low-amplitude recursive loads: case of Groningen gas ﬁeld. In: Proceedings of the
international conference on computational methods in structural dynamics and earthquake
engineering, Rhodes, Greece
3. Manos GC, Kotoulas L, Matsou V, Felekidou O (2015) Dynamic and earthquake behaviour
of greek post-byzantine churches with foundation deformability—experimental investigation
of stone masonry material properties. In: Psycharis I, Pantazopoulou S, Papadrakakis M
(eds) Seismic assessment, behavior and retroﬁt of heritage buildings and monuments.
computational methods in applied sciences, vol 37. Springer
4. Abaqus-Inc. (2013) Abaqus User Manual, Version 6.13. Dassault Systèmes Simulia
Corp. Providence, RI, USA
5. Messali F, Ravenshorst G, Esposito R, Rots J (2017) Large-scale testing program for the
seismic characterization of Dutch masonry walls. In: 16th world conference on earthquake,
Santiago, Chile 
6. Graziotti F, Tomassetti U, Rossi A, Kallioras S, Mandirola M, Penna A, Magenes G (2017)
Full scale shaking table test on a URM cavity wall terraced house building. In: 16th WCEE,
7. Structural Vibration Solutions A/S. ARTeMIS Modal 5 (2016). http://www.svibs.com
9. Draft NPR 9998:2017
10. KNMI (1993) Netherlands seismic and acoustic network. Royal Netherlands Meteorological
Institute (KNMI). Other/Seismic Network. https://doi.org/10.21944/e970fd34-23b9-3411-
11. Smyrou E, Tasiopoulou P, Bal I E, Gazetas G (2011) Ground motions versus geotechnical
and structural damage in the February 2011 christchurch earthquake. Seismological Res Lett
2196 D. Dais et al.