Monitoring, modelling and quantification of accumulation of damage on masonry structures due to recursive loads

The use of induced seismicity is gaining in popularity, particularly in Northern
Europe, as people strive to increase local energy supplies. Τhe local building
stock, comprising mainly of low-rise domestic masonry structures without any
aseismic design, has been found susceptible to these induced tremors. Induced
seismicity is generally characterized by frequent small-to-medium magnitude
earthquakes in which structural and non-structural damage have been reported.
Since the induced earthquakes are caused by third parties liability issues arise
and a damage claim mechanism is activated. Typically, any damage are
evaluated by visual inspections. This damage assessment process has been
found rather cumbersome since visual inspections are laborious, slow and
expensive while the identification of the cause of any light damage is a
challenging task rendering essential the development of a more reliable
approach. The aim of this PhD study is to gain a better understanding of the
monitoring, modelling and quantification of accumulation of damage in masonry
structures due to recursive loads.
Fraeylemaborg, the most emblematic monument in the Groningen region dating
back to the 14 th century, has experienced damage due to the induced seismic
activity in the region in recent years. A novel monitoring approach is proposed to
detect damage accumulation due to induced seismicity on the monument.
Results of the monitoring, in particular the monitoring of the effects of induced
seismic activity,, as well as the usefulness and need of various monitoring data
for similar cases are discussed. A numerical model is developed and calibrated
based on experimental findings and different loading scenarios are compared
with the actual damage patterns observed on the structure.
Vision-based techniques are developed for the detection of damage
accumulation in masonry structures in an attempt to enhance effectiveness of
the inspection process. In particular, an artificial intelligence solution is proposed
for the automatic detection of cracks on masonry structures. A dataset with
photographs from masonry structures is produced containing complex
backgrounds and various crack types and sizes. Moreover, different
convolutional neural networks are evaluated on their efficacy to automatically
detect cracks. Furthermore, computer vision and photogrammetry methods are
considered along with novel invisible markers for monitoring cracks. The
proposed method shifts the marker reflection and its contrast with the
background into the invisible wavelength of light (i.e. to the near-infrared) so that
the markers are not easily distinguishable. The method is thus particularly vi
suitable for monitoring historical buildings where it is important to avoid any
interventions or disruption to the authenticity of the basic fabric of construction..
Further on, the quantification and modelling of damage in masonry structures are
attempted by taking into consideration the initiation and propagation of damage
due to earthquake excitations. The evaluation of damage in masonry structures
due to (induced) earthquakes represents a challenging task. Cumulative damage
due to subsequent ground motions is expected to have an effect on the seismic
capacity of a structure. Crack patterns obtained from experimental campaigns
from the literature are investigated and their correlation with damage propagation
is examined. Discontinuous modelling techniques are able to reliably reproduce
damage initiation and propagation by accounting for residual cracks even for low
intensity loading. Detailed models based on the Distinct Element Method and
Finite Element Model analysis are considered to capture and quantify the
cumulative damage in micro level in masonry subjected to seismic loads.
Finally, an experimental campaign is undertaken to investigate the accumulation
of damage in masonry structure under repetitive load. Six wall specimens
resembling the configuration of a spandrel element are tested under three-point
in-plane bending considering different loading protocols. The walls were
prepared adopting materials and practices followed in the Groningen region.
Different numerical approaches are researched for their efficacy to reproduce the
experimental response and any limitations are highlighted.